refactor: remove unreachable legacy partial-pagination cache-fallback branch in store_or_use_cached_evaluation

[sxxtony]

---

name: Bug Report
about: Report a bug or unexpected behavior
labels: bug

Title

refactor: remove unreachable legacy partial-pagination cache-fallback branch in store_or_use_cached_evaluation

Description

neurons/validator.py:store_or_use_cached_evaluation still contains a legacy branch intended for a pre-mirror partial-pagination failure mode:

if not miner_eval.should_use_cache_fallback and not miner_eval.mirror_pr_fetch_failed:
    bt.logging.warning(
        f'UID {uid}: GitHub fetch failed after partial PR load; skipping cache store/fallback this round'
    )
    continue

After the mirror-only scoring refactor, that state is no longer constructible by current writers of github_pr_fetch_failed.

This leaves stale dead logic and a misleading inline comment in active validator code.

Steps to Reproduce

1. Inspect the stale branch in neurons/validator.py.
2. Check all writers of github_pr_fetch_failed = True:
   • gittensor/validator/oss_contributions/inspections.py
   • gittensor/validator/oss_contributions/mirror/load.py
3. Verify control flow:
   • identity transient path returns early in reward.py before PR load
   • mirror fetch failure sets both mirror/github failure flags and returns before PR append
4. Verify should_use_cache_fallback definition in gittensor/classes.py:
   • github_pr_fetch_failed and total_prs == 0

Expected Behavior

Validator cache-fallback code should not contain legacy branches that are unreachable under the current mirror-only execution model.

Actual Behavior

The legacy partial-pagination branch remains in neurons/validator.py even though:

• only two current writers can set github_pr_fetch_failed = True
• both paths produce total_prs == 0
• one of them also forces mirror_pr_fetch_failed == True

So the guarded condition is unreachable.

Environment

• OS: Ubuntu (VM)
• Python version: N/A (static flow analysis)
• Commit/Version: test branch @ 5291948abe3fd040771c31800b316906e1e92c95 (verified equal across local HEAD, origin/test, and upstream/test)

Additional Context

Unreachable state proof

The stale branch executes only when all are true:

1. github_pr_fetch_failed == True
2. not should_use_cache_fallback == True => total_prs > 0 (given property definition)
3. mirror_pr_fetch_failed == False

But in current code:

• writer A (inspections.py) sets github_pr_fetch_failed only for transient identity failure, and reward.py returns before PR loading -> total_prs == 0
• writer B (mirror/load.py) sets both github_pr_fetch_failed=True and mirror_pr_fetch_failed=True, and returns before PR loading -> total_prs == 0

No third writer exists in gittensor/ or neurons/.

Why this likely remained

• The branch/comment came from hybrid legacy+mirror handling (notably around fix: hybrid-merge cached mirror PRs when only mirror fetch fails #830 context).
• #1202 removed legacy scoring pipeline paths but did not touch neurons/validator.py, so this appears to be incomplete cleanup after large refactor sweep.

Overlap check

• No exact open/closed issue or PR currently targets this specific stale branch cleanup in store_or_use_cached_evaluation.
• Open PR feat(validator): per-repository eligibility #1293 (feat/per-repository eligibility) touches neurons/validator.py but retains this branch.

Proposed fix

Remove only the unreachable block and stale comment in neurons/validator.py, leaving the normal cache-fallback path intact.

Optional follow-up test: assert invariant that any MinerEvaluation with github_pr_fetch_failed=True has total_prs==0 under current writers.