Measure the test suite against #174's targets; replace the Python fakes only if still short #207

Parent

#174

What to build

Check #174's three targets against its baseline, and replace the Python test fakes only if the warm-suite targets are still missed. First measure on an idle machine: warm full cargo test wall time (target ≤ 3 minutes), user + sys (target at most half of the #174 baseline), and a cold cargo test --no-run (target ≤ 5 minutes). If both warm targets are met, post the numbers on #174 and change nothing else. If they aren't, replace the fake gh and claude with something cheaper to start (e.g. a small Rust helper built with the tests) under their current contract: state and records in the JSON files the harness names, the same argv handling and the same outputs. Then measure again.

Acceptance criteria

• A "Result" comment on Make the test suite faster #174 gives all three measurements next to the baseline, with load average and core count
• If the fakes were replaced: every existing scenario passes against the new fakes without changes to the scenarios themselves, and Python is no longer needed to run the suite
• Any target still missed is named in the comment, with where the remaining time goes

Blocked by

• Stop building aws-lc-sys, and axoupdater's image stack if a flag allows #203
• The test harness polls CI about every 100ms, not every 10ms #204
• Unit tests for argument parsing; one Run per --email end-to-end test #205
• Collapse the looping end-to-end scenarios to one Run per behavior #206