Add effective_cache_key(model, context) helper

[ptomecek]

Summary

Adds a public helper effective_cache_key(model, context) to ccflow.evaluators.common.

cache_key(..., effective=True) already derives effective (opt-in) identity, but only for a ModelEvaluationContext. Callers that hold a bare model plus a context — for example keyed persistent caches — otherwise have to build a throwaway ModelEvaluationContext and then compare the structural and effective keys to detect opt-out models. This helper wraps the existing internal derivation directly:

def effective_cache_key(model: CallableModel, context: ContextBase) -> bytes:
    try:
        key = _effective_model_key(model, context, {}, set())
    except _EffectiveEvaluationKeyUnavailable:
        key = None
    return key if key is not None else cache_key(model)

• Models that declare an effective identity (non-None identity payload) are keyed by it.
• Models that do not opt in fall back byte-for-byte to cache_key(model).
• Unlike cache_key(mec, effective=True), the result is the model-level effective key without the surrounding evaluation-context envelope (fn/options), and opt-out models get the context-independent structural key rather than a context-dependent one.

No behavior change to existing APIs — this is purely additive (new public function + __all__ entry).

Tests

TestEffectiveCacheKey covers: opt-out equals cache_key(model) and is context-independent; opt-in is keyed by the identity payload (and collapses contexts the payload treats as equivalent); payload-relevant differences produce distinct keys; and errors raised inside an identity payload propagate rather than being masked by the structural fallback.

ruff check clean; ccflow/tests/evaluators/test_common.py (45) and the effective-key characterization suite pass.

[github-actions]

Test Results

    1 files  ±0      1 suites  ±0   2m 32s ⏱️ -29s
1 352 tests +5  1 350 ✅ +5  2 💤 ±0  0 ❌ ±0 
1 358 runs  +5  1 356 ✅ +5  2 💤 ±0  0 ❌ ±0 

Results for commit 6b652de. ± Comparison against base commit a143162.

♻️ This comment has been updated with latest results.

[codecov]

Codecov Report

❌ Patch coverage is 97.05882% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 93.53%. Comparing base (a143162) to head (6b652de).

Files with missing lines	Patch %	Lines
ccflow/evaluators/common.py	83.33%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #265   +/-   ##
=======================================
  Coverage   93.53%   93.53%           
=======================================
  Files         176      176           
  Lines       20460    20494   +34     
  Branches     1352     1352           
=======================================
+ Hits        19137    19170   +33     
- Misses       1052     1053    +1     
  Partials      271      271           

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

🚀 New features to boost your workflow:

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