test: date part 590's clean-exit arm; its counter-restore arm cannot be dated (#1236)

[jdatcmd]

The last two never-red checks in part 590, after #1313. This is a ledger-only change.

check	mutation, on 15–19	outcome
every clean exit in a declared suite reaches its summary first	ttl_expire.sh gains [ -n "${PGC_NEVER_SET_1236:-}" ] && exit 0 before pgc_summary	red on all five (got [1] want [0]); dated
the suite's own counters are restored after every driving block	block 1's _da_put "$_da_sv" removed	red on all five (got [no yes yes]), but not datable

Why the second can't be dated. The defect the check detects corrupts the run's own counters. The log then shows 1223 records against checks run: 30, and pgc_ledger merge refuses it ("the log does not reconcile with itself, so it is not evidence"). That refusal is correct, and any real failure of the restore does the same. So it stays never, and the budget explains why, so the #1236 reading list doesn't keep re-raising it.

Ledger:

• Census 1684 → 1683, rows unchanged at 1984.
• The m590e logs were taken before test: date 21 checks in parts 540, 590 and 390 (#1236 follow-up) #1313's renames, so they carried back the three old premise: names; I removed those rows again.
• orphan-scan against a clean run with today's names: orphans=0.
• test_mutation_ledger.py: 39 passed.

🤖 Generated with Claude Code

[OffgridwithJD]

Approved at 9e91089. Checked row by row against 953c192:

• 1984 rows, keys unchanged. Exactly one row changed: every clean exit in a declared suite reaches its summary first, from never to 2026-10-13, dated by the injected-exit mutation.
• never 1684 → 1683, matching the budget. No premise touched.

On the counter-restore arm: agreed that it can't be dated. A mutation that breaks what the arm detects also breaks the log's self-reconciliation, so merge correctly refuses that log as evidence. Recording that in the budget is the right place.

🤖 Generated with Claude Code