[Enhancement] Split cast/agg template-cartesian TUs to cut compile time

[Hyper-FF]

Why I'm doing:

Several expression/aggregate factory TUs were dominated by large 2-D template cartesians (arg-type x ret-type / from-type x to-type), making them among the slowest single translation units in the BE build. Split each, behavior-preservingly, across parallel TUs: move the classes/templates/macros into a shared _tpl.hpp and distribute the cartesian over group .cpp files, with the original file routing by one type dimension.

• aggregate_resolver_avg: split out sum_map -> aggregate_resolver_sum_map{1,2,3}.cpp (641s -> 201 + 150/154/117s)
• case_expr: result-type cartesian -> case_expr_g{1,2,3}.cpp (536s -> 14 + 203/176/157s)
• aggregate_resolver_minmaxany: split out max_by/min_by ->
  aggregate_resolver_maxminby{1,2,3}.cpp (346s -> 187 + 90/90/76s)
• cast_expr: primitive from-type x to-type cartesian -> cast_expr_g{1,2,3}.cpp (258s -> 79 + 87/88/72s)

Behavior-preserving (identical type coverage). The wall-clock pole of each TU drops ~60%; total CPU rises slightly from per-group header re-parse. Isolated compile times measured without PCH.

What I'm doing:

Fixes #issue

What type of PR is this:

• BugFix
• Feature
• Enhancement
• Refactor
• UT
• Doc
• Tool

Does this PR entail a change in behavior?

• Yes, this PR will result in a change in behavior.
• No, this PR will not result in a change in behavior.

If yes, please specify the type of change:

• Interface/UI changes: syntax, type conversion, expression evaluation, display information
• Parameter changes: default values, similar parameters but with different default values
• Policy changes: use new policy to replace old one, functionality automatically enabled
• Feature removed
• Miscellaneous: upgrade & downgrade compatibility, etc.

Checklist:

• I have added test cases for my bug fix or my new feature
• This pr needs user documentation (for new or modified features or behaviors)
  • I have added documentation for my new feature or new function
  • This pr needs auto generate documentation
• This is a backport pr

Bugfix cherry-pick branch check:

• I have checked the version labels which the pr will be auto-backported to the target branch
  • 4.1
  • 4.0
  • 3.5

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 2 committers have signed the CLA.

✅ wanpengfei-git
❌ Hyper-FF
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[github-actions]

Module-risk briefing for Codex review

This is review focus, not a finding list. Validate against the diff.

High-confidence risks

• Module: exprs/agg
  Changed files: be/src/exprs/agg/factory/aggregate_factory.cpp, aggregate_resolver.hpp, aggregate_resolver_avg.cpp, aggregate_resolver_maxminby.hpp, aggregate_resolver_maxminby{1,2,3}.cpp, aggregate_resolver_minmaxany.cpp, aggregate_resolver_sum_map.hpp, aggregate_resolver_sum_map{1,2,3}.cpp
  Historical failure pattern: registration/lifecycle and type-coverage gaps introduced when aggregate resolver code is split, moved, or ported across TUs (missing track_function() wrapper, cache-key/arg-count mismatches, template-parameter mismatches breaking builds).
  Review focus: this PR splits register_avg's sum_map cartesian into register_sum_map{1,2,3}() and register_minmaxany's max_by/min_by cartesian into register_maxminby{1,2,3}(). Confirm the three key-type/outer-type partitions in each split are an exact, non-overlapping, non-omitting partition of the original combined type list (sum_map key types across sum_map1/2/3; max_by/min_by outer types across maxminby1/2/3), and that all six new register_* functions are actually invoked from AggregateFuncResolver::AggregateFuncResolver() (they are, in aggregate_factory.cpp) — a silently-dropped partition or unregistered function would be indistinguishable from working code until a specific type combination is queried.
  Evidence: [BugFix] fix clang build failure in template parameter #70932, [BugFix] fix asan crash #67678, [BugFix] Fix memory leak in count aggregate function registration #67490, [BugFix] vararg Udaf's cache key should contains num_args #72102 (from 19 merged bugfixes)
• Module: exprs
  Changed files: be/src/exprs/case_expr.cpp, case_expr.h, case_expr_g{1,2,3}.cpp, case_expr_tpl.hpp, cast_expr.cpp, cast_expr_g{1,2,3}.cpp, cast_expr_tpl.hpp
  Historical failure pattern: VectorizedCaseExpr::generate_ir_impl previously selected the LLVM compare instruction by ResultType instead of WhenType, producing malformed JIT IR when the two types diverge in float-ness (PR [BugFix] Fix invalid JIT IR for CASE WHEN with mixed float/int when and result types #74382).
  Review focus: case_expr_tpl.hpp moves generate_ir_impl (and the WhenType/ResultType-dependent JIT codegen branch) verbatim out of case_expr.cpp — verify the WhenType-based branch selection from the [BugFix] Fix invalid JIT IR for CASE WHEN with mixed float/int when and result types #74382 fix survived the move intact and wasn't accidentally reverted to ResultType. Also verify the three-way result-type partitions in case_expr_g{1,2,3}.cpp and cast_expr_g{1,2,3}.cpp cover every LogicalType the original single-file switch covered exactly once (no type silently dropped, causing create_primitive_cast/case_from_thrift to return nullptr for a previously-supported type), and that the routing in case_expr.cpp/cast_expr.cpp correctly chains to all three groups.
  Evidence: [BugFix] Fix invalid JIT IR for CASE WHEN with mixed float/int when and result types #74382 (from 20 merged bugfixes)

[CelerData-Reviewer]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Can't wait for the next one!

Reviewed commit: 530ad6a726

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[github-actions]

[FE Incremental Coverage Report]

✅ pass : 0 / 0 (0%)

[github-actions]

[Java-Extensions Incremental Coverage Report]

✅ pass : 0 / 0 (0%)

[github-actions]

[BE Incremental Coverage Report]

❌ fail : 1059 / 1358 (77.98%)

file detail

	path	covered_line	new_line	coverage	not_covered_line_detail
🔵	be/src/exprs/case_expr_g3.cpp	3	11	27.27%	[23, 24, 25, 26, 27, 29, 30, 31]
🔵	be/src/exprs/case_expr_g2.cpp	6	11	54.55%	[23, 25, 27, 28, 30]
🔵	be/src/exprs/case_expr_g1.cpp	6	11	54.55%	[24, 27, 28, 29, 30]
🔵	be/src/exprs/cast_expr_tpl.hpp	614	843	72.84%	[144, 152, 153, 197, 202, 208, 209, 210, 213, 215, 217, 232, 234, 235, 236, 241, 242, 243, 244, 269, 270, 271, 293, 294, 295, 296, 299, 300, 306, 307, 308, 326, 327, 332, 333, 338, 340, 341, 342, 344, 356, 357, 359, 362, 382, 383, 384, 404, 405, 406, 425, 426, 432, 433, 434, 452, 453, 463, 464, 474, 488, 491, 492, 493, 517, 525, 526, 545, 546, 547, 564, 565, 566, 596, 597, 612, 615, 616, 617, 618, 629, 632, 633, 634, 635, 637, 646, 656, 658, 676, 680, 681, 683, 689, 700, 704, 705, 706, 707, 709, 710, 714, 715, 724, 728, 729, 730, 731, 738, 752, 753, 754, 755, 764, 765, 768, 769, 770, 774, 775, 776, 784, 785, 787, 788, 791, 792, 793, 794, 795, 796, 797, 798, 799, 800, 802, 816, 817, 818, 834, 835, 836, 854, 855, 863, 864, 865, 879, 880, 881, 883, 884, 885, 886, 887, 903, 904, 905, 921, 922, 923, 955, 956, 970, 972, 973, 976, 977, 987, 999, 1000, 1024, 1025, 1026, 1043, 1044, 1045, 1068, 1076, 1077, 1079, 1080, 1082, 1083, 1084, 1085, 1120, 1121, 1132, 1133, 1134, 1135, 1144, 1145, 1146, 1147, 1154, 1155, 1165, 1166, 1171, 1172, 1215, 1216, 1222, 1228, 1229, 1249, 1250, 1291, 1293, 1295, 1353, 1355, 1358, 1360, 1405, 1406, 1407, 1408, 1506, 1508, 1514, 1520, 1541, 1542, 1586, 1602, 1603]
🔵	be/src/exprs/case_expr_tpl.hpp	286	338	84.62%	[60, 72, 100, 119, 124, 125, 135, 159, 169, 173, 174, 188, 189, 191, 193, 194, 195, 196, 197, 199, 201, 203, 204, 205, 206, 207, 216, 217, 218, 228, 240, 258, 259, 260, 261, 262, 263, 264, 265, 267, 269, 270, 272, 273, 275, 278, 279, 281, 282, 283, 387, 555]
🔵	be/src/exprs/case_expr.cpp	10	10	100.00%	[]
🔵	be/src/exprs/cast_expr_g3.cpp	6	6	100.00%	[]
🔵	be/src/exprs/agg/factory/aggregate_resolver_maxminby.hpp	13	13	100.00%	[]
🔵	be/src/exprs/cast_expr_g2.cpp	8	8	100.00%	[]
🔵	be/src/exprs/agg/factory/aggregate_resolver_sum_map1.cpp	3	3	100.00%	[]
🔵	be/src/exprs/agg/factory/aggregate_resolver_maxminby3.cpp	14	14	100.00%	[]
🔵	be/src/exprs/agg/factory/aggregate_resolver_maxminby2.cpp	15	15	100.00%	[]
🔵	be/src/exprs/cast_expr_g1.cpp	8	8	100.00%	[]
🔵	be/src/exprs/cast_expr.cpp	19	19	100.00%	[]
🔵	be/src/exprs/agg/factory/aggregate_resolver_sum_map2.cpp	3	3	100.00%	[]
🔵	be/src/exprs/agg/factory/aggregate_resolver_maxminby1.cpp	15	15	100.00%	[]
🔵	be/src/exprs/agg/factory/aggregate_resolver_sum_map.hpp	21	21	100.00%	[]
🔵	be/src/exprs/agg/factory/aggregate_resolver_sum_map3.cpp	3	3	100.00%	[]
🔵	be/src/exprs/agg/factory/aggregate_factory.cpp	6	6	100.00%	[]