refactor: remove obsolete Discovery and Messaging paths

[jeremi]

Change

Remove Discovery's deprecated Node/Python structural-validation aliases and their generated facade exports. Adopters use validateSelectionStructure or validate_selection_structure, then explicitly accept a selection through local policy.

Fresh Messaging schemas contain only the shared activation ledger. Remove the unused legacy table creation and cover the resulting schema with a real PostgreSQL regression. This is a fresh baseline change, with no legacy-state migration.

Verification

• Messaging: seven PostgreSQL migration tests and thirteen activation/package tests passed, including concurrency and split-role authority; focused strict Clippy and contract checks passed.
• Discovery: rebuilt Node (9) and Python (14) tests, generated declarations, contract/HTTP checks and native handoff journey passed. The Node smoke used a separately installed local tarball; Python used the rebuilt extension.
• Independent review found no substantive issue. After the latest base synchronization, facade sync, all three facade tests and TypeScript checks pass. The reviewed 22-file feature diff remains byte-identical.

Current pushed head: 4e88d0e540d699ce8fd5c1a45adf432cd6872601, based on PR #2040 at 6965700ae483975a0b6678c5c0431555577e34c2. Review against jeremi/product-cleanup-integration. The signed-off base synchronization preserves the exact reviewed 22-file feature diff; facade sync, all three facade tests and TypeScript pass.

The maintained manual CI workflow passed on this exact synchronized head with normal changed-path selection: 54 successful jobs and three skipped. Automatic CI does not select this stacked base. The synchronized base includes independently accepted BReg Review identity and Coordinator terminal-evidence corrections, preserving the legitimate restore-review case. Final base proof passes 229 BReg, 42 Coordinator and ten Rust facade tests, a real PostgreSQL recovery/retention regression, and freshly rebuilt installed public tests (108 Node, 127 Python). Earlier fixture and interrupted-S3 corrections remain included. Independent feature and base-correction review are complete. The initial automated review failed on f28 and supplies no automated clearance; base PR #2040's automatic review completed on its current head at 14:12Z without new findings. Current-head CI is complete. Base PR #2040 is also green and conflict-free, but its 264 commits exceed the configured rebase queue's 100-commit limit; that merge-method decision remains open.

No release, deployment, external publication or main-branch merge is included.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	⚠️ Failed 2026-10-15T05:14:25.889194Z	f28e0e8	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.