Move ReferencedClassFinder to base

[johnynek]

Ideally anything that does not increase the dependencies should be moved to base. By removing a function it looks like no one was using (and is just a trivial redirection to Config) we can do that here (although the tests do reference things in code).

cc @navinvishy @daniel-sudz I think we could use this code in the beam and spark backends in each stage of writes to try to create tokens for all the classes we can see. In kryo, this is registering classes so they have names. Note, I would sort the classes first so we assign the classes in a reproducible way that never depends on any iteration order of a hash set.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 37.78%. Comparing base (6434348) to head (fe647f8).
⚠️ Report is 6 commits behind head on develop.

Additional details and impacted files

@@              Coverage Diff              @@
##             develop    #1987      +/-   ##
=============================================
- Coverage      37.82%   37.78%   -0.05%     
- Complexity      1116     1335     +219     
=============================================
  Files            316      315       -1     
  Lines          21117    21061      -56     
  Branches        2924     2874      -50     
=============================================
- Hits            7988     7957      -31     
+ Misses         12141    12123      -18     
+ Partials         988      981       -7     

see 1 file with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6434348...fe647f8. Read the comment docs.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[navinvishy]

In the beam backend, I see that we're getting a kryo instantiator here. We're using KryoHadoop by default which gets Cascading serialization tokens and registers those classes with Kryo. Are you suggesting removing the dependency on KryoHadoop?

[hazel-sudz]

this change seems fine to me

I don't quite understand this part though:

I think we could use this code in the beam and spark backends in each stage of writes to try to create tokens for all the classes we can see. In kryo, this is registering classes so they have names. Note, I would sort the classes first so we assign the classes in a reproducible way that never depends on any iteration order of a hash set.

what is the use case for this? Having writers be aware of all exiting classes?