[findalexli]

Task finding: #42. Merge after #43, which removes the crash this finding reports; until then the solver in #42 would go from a valid 0 to invalid.

What and why

When evaluate.py crashes, parse_result.py finds no score file and reports valid: true, score 0, so a judge failure looks like a solver that scored nothing. run.sh now passes on the evaluator's exit status, and no score file with a status other than 0 or 2 gives valid: false. evaluate.py's "solver is not executable" exit changes from 1 to 2, so a missing solve stays a valid 0, as today. This looks only at a missing score file, so Known issue 1 stays.

Old vs new judge

Old key cd3a5bf1f538, new 3ef6c9298492; sforge eval, one worker, order swapped, n=2 per judge. The new image is the published old tag plus this branch's judge files, because two full builds don't fit a worker's disk.

Submission	Old	New
The reproducer in #42, evaluator crashes	0, valid, correct=0/0	Invalid, "evaluator exited with status 1 before writing a score file"
Fails to build, or no solve, or rejected by preflight	0, valid, correct=0/0	Same
Always unknown, or wrong models	0, valid, correct=0/774	Same
Replay fixture, correct models on 42 files	6.28, correct=42/774	Same

A valid 0 from a crashed evaluation has no score to map; regrade it. No other score moves.

Checklist for task changes

• version in task.toml is bumped (any change that can move scores). If smt_solver: validate deep models on a large stack #43 merges first, this becomes 1.2.0.
• NOTES.md: the change and its score effect are under Changes. No Known issue covered this.
• Old vs new judge on the same submissions, on the same host, results above.
• New blobs: none.