Render one pitch gap without 1× prefix

[pzfreo]

Summary

A two-member pattern has one pitch gap. Render its approved pitch as 650, not 1× 650. Multi-gap spans keep their multiplier (3× 25). The shared renderer covers hole, slot and pocket patterns; the independent pocket-pattern drawing evaluator now expects the same honest presentation while checking the confirmed measurement identity.

Evidence

• The same AP242 CTC01 fixture in both modes now labels its four single-gap pitches 90, 320, 350, and 650. A generated preview PDF was rendered and inspected.
• 66 focused hole, slot, pocket and CTC01 tests passed. Restoring the old 1× formatter deliberately failed test_grid_pattern_renders_both_pitch_dims; the fix was restored byte-for-byte afterward.
• scripts/pr-check --static passed on the final change.
• On the automatic CTC01 PMI build, the before/after lint totals stayed at 5 errors, 5 warnings and 12 infos; PMI stayed at 21 rendered and 6 dropped. This formatting fix does not claim to resolve that sheet's existing incompleteness.

Architecture

The approved pitch value and measurement identity still come from the ADR 1/4 compiled plan. Only the render text for a single gap changes. ADR 2 placement uses the same endpoints and shorter measured label. ADR 3 recognition is unchanged. ADR 5's independent drawing evaluator still checks the visible label against the confirmed claim. No ADR boundary changes.

The separate CTC01 repeated nominal diameter and distinct member-specific PMI presentation is tracked in #2172; this PR does not suppress either claim.

Follow-up at 785968c

The stacked #2174 full matrix found two additional consumers of the intentional one-gap text change. The grid-plate exact golden now records 20 and its actual label bounds. The independent hole-pattern drawing evaluator now accepts a bare value for one interval and still requires the correct explicit multiplier for longer spans. Its mutation test proves that changing the real one-gap 20 to 9× 20 loses drawing credit; a wrong nominal also loses credit.

The evaluator and golden updates passed 15 focused hole-pattern tests, all 23 refactor-golden cases, the exact grid-plate golden, ruff, mypy, and the source-complexity guard locally. A fresh independent review checked the final parent diff and its merge into #2174, found no concrete defect, and verified ADR 1–5 fit. The current exact-head CI is the merge gate.