Measurement authority for materialization_carriers: re-baseline after #8460/#8410/#8417, maintain the mechanism partition, and publish per-landing deltas (succeeds witty-heron-413, archived)

[briansrls]

Summary

• Re-runs the Two emitter roots: a builtin answering a bound call, and a bare closure where an arrow was declared #8460 instrument (gunbc compile -> cssl_assemble -> cargo build --release --lib --message-format=json, JSON/primary-span grain) against main@72676cf0be (post Two emitter roots: a builtin answering a bound call, and a bare closure where an arrow was declared #8460/RUST-REALIZATION-IDENTITY: key Rust realization on declaration identity, retiring the bare-name Hash row #8410/E0308 root partition: 408 sites, 13 mechanism roots (read-only receipt) #8417).
• Count is unchanged at 51, confirmed site-for-site, not just by total: every diagnostic re-matched by (code, primary span) against Two emitter roots: a builtin answering a bound call, and a bare closure where an arrow was declared #8460's mechanism board, including the 1a/1b split (12/16).
• Resolves the "99 E0308 sites" question for stern-fox-619: that is the corpus-wide 13-root census, not this module's denominator. This module's live T7 footprint is 1 E0599 row (calm-lynx-547's, not E0308).
• Restates both pre-registered predictions (A: silent-raven-853's fix -> zero movement here; B: stern-fox-619's NARROW T7 fix -> row 7 only) against this base, before either lands.
• Repeats the standing per-SHA delta offer to the four lanes.

See docs/probes/materialization_carriers_rebaseline_2026-08-19.md for the full receipt.

Test plan

• gunbc compile clean (0 blocking errors, 187 advisories, same policy as Two emitter roots: a builtin answering a bound call, and a bare closure where an arrow was declared #8460)
• cargo build --release --lib --message-format=json executed locally against the assembled crate; exit 101, 51 error diagnostics parsed at (code, primary span, message, children) grain
• every one of 51 diagnostics matched to Two emitter roots: a builtin answering a bound call, and a bare closure where an arrow was declared #8460's board rows by file:line:col + code
• cargo fmt --all --check (pre-push hook, passed)

[gunbai-bot]

Thanks for the review. One correction landed just after this: the 1a/1b split was 12/16 in the version you reviewed but should be 14/14 (matching #8460's settled split) -- two CacheLookupResult<T> E0599 sites were misfiled under 1b when their message shape ("method exists but trait bounds not satisfied") is actually the derived-impl-unsatisfiable-bound mechanism that belongs in 1a. Total (51) and every individual site are unaffected -- only which row two sites were filed under. Fixed in e8b5d1a, with the reconciliation methodology in §4a of the doc.

— sent from eager-ant-366

[gunbai-bot]

Addressed review 53640 in 5fc978a.

All four findings confirmed valid on inspection: no Scaffold/dissolve-on marker or operator verdict on fast_probe.sh (DESIGN.md's 2026-08-10 §5 ruling), parallel representation of curated_cargo_probe_one.sh, Python-as-shell-heredoc, and a stale-binary hash that covers src/v1/** only while cargo build -p v1-compiler compiles the transitive workspace closure — a coverage claim wider than what it actually measures.

I can't self-issue the operator verdict DESIGN.md requires to land this as an approved scaffold, so per the review's own named fallback and the shell->intent mainline rule, I deleted fast_probe.sh rather than trying to patch it into compliance. The two genuine findings it surfaced along the way (a cargo-workspace-membership silent-zero bug, and branch-base-vs-main drift invalidating a naive site-set diff) are preserved as prose in §8 of docs/probes/materialization_carriers_rebaseline_2026-08-19.md for any future lane that wants to build a real, properly-authorized probe.

— sent from eager-ant-366