[release/cvs-0.2.0] cvs/rccl: fix dmesg false positives and per-user RCCL hostfile

[speriaswamy-amd]

Summary

Cherry-pick of #169 onto release/cvs-0.2.0. Three small fixes from debugging a 28-node rccl_perf run on a shared cluster:

• cvs/lib/verify_lib.py — surface amdgpu runlist oversubscription as a non-fatal WARN. Per AMD docs, this is a real perf-degrading state but the collective itself completes correctly. New module-level warn_patterns_dict scanned alongside err_patterns_dict inside verify_dmesg_for_errors. Backward-compatible signature & return shape; existing callers untouched.
• cvs/tests/rccl/rccl_perf.py — verify_dmesg_for_errors(..., till_end_flag=False) so each parametrized test's dmesg scan is bounded by its own start/end window (prevents one test's kernel event from cascading into N reported failures).
• cvs/lib/rccl_lib.py — per-$USER mpirun hostfile path (/tmp/rccl_hosts_file_<USER>.txt); replaces the sudo rm -f workaround that was on this branch.

Conflict resolution notes

• cvs/lib/rccl_lib.py: this branch had a sudo rm -f /tmp/rccl_hosts_file.txt workaround for the same multi-user collision. The per-user path supersedes it; workaround removed.
• cvs/tests/rccl/rccl_perf.py: main gates verify_dmesg_for_errors on if can_use_sudo: (passwordless-sudo guard not yet on this branch). For this cherry-pick the call stays unconditional — only the till_end_flag value is changed — to avoid introducing an undefined can_use_sudo reference.
• cvs/lib/verify_lib.py: auto-merged cleanly.

Test plan

• Cherry-pick produces the same final diff as cvs/rccl: surface oversubscription as WARN, per-test dmesg window, per-user hostfile #169 on main.
• python -c "import ast; ast.parse(...)" on all three files passes.
• Smoke-run rccl_perf from this branch on the validation cluster.

Refs: AIMVT-175

[cijohnson]

please add links to the reference doc in the comment for future reference