claw: mutation sweep of check_state.t27 and merge_lane.t27, 15 real survivors #10724

What

A mutation sweep of two TRI CLAW law specs (mutgen.sh operators, t27c test-report per mutant on the Railway lab):

• specs/claw/check_state.t27 reads a PR's check rollup into pass / failing / pending / unreadable and a verdict.
• specs/claw/merge_lane.t27 measures the auto-merger: lane state, yield and its share of the queue.

Real survivors (a test is missing)

check_state.t27:

• 70 or -> and: PENDING and REQUESTED stop being wait statuses. No test reads either one.
• 139 rows < CHECK_MAX_ROWS -> <=: the 1024-row cap has no test at all.
• 158 ls = le + 1 -> + 2 in dashboard_failing: the second row loses its first byte, so a CONTEXT row is read as ONTEXT and counted failing.

merge_lane.t27:

• 125, 217, 330 le > ls -> >=: a blank line becomes an unreadable row in ml_counts, my_counts and ms_counts.
• 128, 219, 332 rows < ML_MAX_ROWS -> <=: none of the three 4096-row caps has a test.
• 130 age >= 0 -> >: an open run of age 0 becomes unreadable.
• 229 nl > 0 -> >=: an empty run with no missing label is counted as label-only.
• 308 e <= s -> < and return false -> return true: an empty event name is accepted.
• 312 b >= 'a' -> > and b <= 'z' -> <: the letters a and z are refused in an event name. No test event contains an a (workflow_run, schedule and pull_request have none).

Equivalent (cannot change any output)

• ls < n -> <= (check_state 135, 153; merge_lane 123, 215, 328): at ls == n, qc_line_end returns n, so the empty range is skipped.
• check_state 155 le > ls -> >= in dashboard_failing: an empty range never splits, so it is not counted either way.
• merge_lane 134 > -> >= on the oldest age: an equal age writes the same value.
• merge_lane 173, 274 rows > 0 -> >=: with no rows, the verdict from all-zero counts is already idle and none.
• check_state 205: the t_class test helper.

Also in this change

#10722: the CR tests in both specs wrote "\r", which is a backslash and an r in a t27 literal. The tests now write the byte 13 and assert that it is there.

Boundary

• specs/claw/check_state.t27
• specs/claw/merge_lane.t27
• .trinity/seals/claw_ClawCheckState.json
• .trinity/seals/claw_ClawMergeLane.json

User Scenarios

• Given a mutant of specs/claw/check_state.t27 or specs/claw/merge_lane.t27 listed above as a real survivor, when t27c test-report runs on it, then at least one test fails.
• Given an equivalent mutant, when it survives, then this issue records why it cannot change any output.

Requirements

• FR-001: Production code MUST stay unchanged; only tests are added.
• FR-002: Each of the 15 real survivors MUST be caught after the change.
• FR-003: The carriage-return tests MUST write the byte 13 and assert that it is present (TRI CLAW: two claw tests write CR as "\r" in a literal, which is a backslash and r #10722).

Success Criteria

• check_state reports 17 tests passed and merge_lane 30 tests passed under t27c and t27b.
• The full re-sweep catches 36 of 40 and 86 of 92 mutants.
• seal --verify exits with exit 0 for 32 files.

Status

Done by PR #10726 (merged).