[squid-protocol]

Part of #7901, #7709. Checklist: done [c:7124add3]

What

The T9 re-survey on the 39 basis (owner ruling 2026-11-06, #7709 decision 3a), plus the small tool change that lets the survey say which basis it used.

• tests/tools/cics_census.py: blockers (text and JSON) and the history line carry copy_members and print a basis: line naming the copy members listed and not counted. The key is added only when non-empty.
• Tests added in test_cics_census_copy_member_7901.py for both the present and absent cases.
• Re-survey of origin/main 7dc1d0138 (census clones, tools only, aggregates, no census source read or copied): blockers_history.jsonl line, regenerated gap_prevalence.md and census_summary.json.

T9 row (counts only)

Census half: 11 of 39 translate whole, 35 of 39 accounted. The 40th member is a copy member, listed and not counted. Ceiling under the rulings: 37 of 39.
Coverage if closed: top 1 -> 12, top 3 -> 23, top 5 -> 31, top 10 -> 36 (upper bounds). Burned half: 126 of 129, unchanged.
Accounted moved 36 -> 35 and top-1 13 -> 12 against the page #7926 committed. The WEB rows now read WEB CONVERSE POST as one gap, one program moved from undeclared item to other, and the survey ran on a later main. Not attributed further here.

Class / Sibling gaps

Class: a census report that counts programs but hides the copy members it excluded. Handled here: blockers text and JSON, the history line. Already handled: gap_prevalence page (#7926), unmask report, survey.json rows. The nightly publish reads blockers.json, so it picks the field up. Sibling gaps: none. Burned estates only needed.

The row is posted on #7709 (comment) with the basis change named.

🤖 Generated with Claude Code

[squid-shepherd]

No description provided.

[squid-shepherd]

No description provided.

[squid-shepherd]

This PR conflicts with main, so no checks ran. Merge origin/main into it and regenerate generated files with their tools (never hand-merge them), then push.

[squid-shepherd]

CI shepherd: the head of this PR (bddb5b9) is not the one a maintainer labelled shepherd:merge on (new commits, or never approved), so the label was removed and nothing merges unreviewed code. Re-add the label when the new head is ready (shepherd:reapprove marks this PR until then). Until then it is a draft and runs the cheap audits only; re-adding the label marks it ready, which starts the full set on that head.

[squid-shepherd]

No description provided.

[squid-shepherd]

No description provided.