test(train): the models row test pins the row #5142 made, a dispatch is a night

[noahgift]

LAB fix for #5165: the nightly train's own case table read 10/13 because scripts/tests/models_lane_row_test.sh still pinned the models row from before #5142.

#5142 made a dispatched run on main count as the night, and changed the models row's event pattern from ^schedule$ to ^(schedule|workflow_dispatch)$. This PR updates the test to that rule:

• the byte pin of the row (constant and header);
• planted_dispatch_never_reaches_verdict becomes newer_dispatch_red_is_the_night: a dispatched red newer than the scheduled green is read, and is red;
• dispatch_alone_is_not_measured becomes dispatch_alone_is_the_night: green, from that run;
• mutant m2 is turned around to plant the old schedule-only pattern; two rows kill it.

Result: 13/13 rows, 5/5 mutants killed. Test-only change; LAB, not a merge or release gate.

Closes #5165
Refs: #5142
keep-open: #5142 -- merged PR cited as the source of the rule; nothing to close

ont-delta: none -- test-only change; no ontology term changes

🤖 Generated with Claude Code

[github-actions]

§13.11 rung 1 — quorum shadow verdict

S13-SHADOW pr=5166 head=ec912e8afbd83a6027d0cc37107dec74b1e30b17 verdict=REFUSE class=Q1 arm_rc=1

Shadow mode: this records a verdict and merges nothing. A refusal
to arm is not a block (§13 adds zero rows to §7) — the pull request is
exactly as green as it was.

[noahgift]

Seat 2, aprender-w4664. Head ec912e8. VERDICT: APPROVE (posted as a comment, because a shared account cannot approve; void on any push)

Run on this head:

• scripts/tests/models_lane_row_test.sh: 13/13 rows pass, rc 0. That includes newer_dispatch_red_is_the_night (red, run 503) and dispatch_alone_is_the_night (green, run 503).
• scripts/tests/models_lane_row_test.sh --mutants: 5/5 mutants killed, rc 0. The new m2 plants the old schedule-only event pattern, and 2 rows break on it.
• The pinned ROW is byte-identical to the live row at scripts/release/nightly_train.sh:112, which is ^(schedule|workflow_dispatch)$. So the test now pins the rule #5142 shipped, not the one that was there before it.
• Main's check_pr_closes_issue.sh passes on the PR body.

Read: the change is test-only and the header table matches the new behaviour rows. The rename and inversion of the two dispatch rows follow from #5142's rule. I have no findings.

[noahgift]

seat 1, aprender-78 APPROVE @ ec912e8

• The new byte pin matches the models row the nightly train carries on main: ^(schedule|workflow_dispatch)$;^models$.
• RED before: the parent's test, run against this tree, gives 10/13. row_is_pinned, planted_dispatch_never_reaches_verdict and dispatch_alone_is_not_measured break.
• GREEN at this head: 13/13 rows, rc 0. --mutants kills 5/5; the inverted m2 (the old schedule-only row) breaks 2 rows.
• This is a test-only diff, 8 lines in one file. The two renamed rows assert the rule #5142 set: a dispatched run on main counts as the night. No other row or threshold changes.
• CI on this head: 10 pass, 2 skipped.