chore: Misc basic evals improvements (eg: consistent activation heuristic, cli args)

[jbloomAus]

Description

Updates to the evals code to compute feature-wise metrics, adds tests and improves CLI interface.

Type of change

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

Checklist:

• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I have not rewritten tests relating to key interfaces which would affect backward compatibility

[hijohnnylin]

@jbloomAus does this require us to rerun any evals? also, has the schema changed? (if so we need to update on np's side)

[codecov]

Codecov Report

Attention: Patch coverage is 73.94958% with 31 lines in your changes missing coverage. Please review.

Project coverage is 64.13%. Comparing base (36e1d86) to head (d1b4f5d).
Report is 27 commits behind head on main.

Files with missing lines	Patch %	Lines
sae_lens/evals.py	74.57%	23 Missing and 7 partials ⚠️
sae_lens/training/sae_trainer.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #324      +/-   ##
==========================================
+ Coverage   63.97%   64.13%   +0.15%     
==========================================
  Files          25       25              
  Lines        3223     3318      +95     
  Branches      408      421      +13     
==========================================
+ Hits         2062     2128      +66     
- Misses       1052     1074      +22     
- Partials      109      116       +7     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.