chore(release): prepare MCP 1.0.23

[escott-]

Bumps the workspace, npm package, MCP Registry metadata and the boundary self-test to 1.0.23, and adds the 1.0.23 changelog.

What ships (4 commits since v1.0.22):

• Agent Coach Kit compare marks feedback its job kept for that job only, and says when it could not check a scope; the coach card meets WCAG 2.1 AA (22be1ee)
• Codex Cloud kit: carried items, verdicts and metrics (5a383d4, cd77a8f), repository only
• Agent Coach Kit package for the ChatGPT and Codex plugin directory in plugins/agent-coach-kit (e0bfb25), repository only; nothing submitted

No dependency changes.

Checked on ovh-desktop against main e0bfb25 plus this change:

• cargo fmt --all --check; cargo clippy --locked --workspace --all-targets -- -D warnings
• cargo test --locked --workspace --all-targets: 3278 passed, 0 failed, 21 ignored
• account_connection_smoke: 9 tests
• cargo audit --deny warnings (403 crates, no advisories) and cargo deny --locked check
• On a clean archive: release_contract metadata and resolve-version for tag v1.0.23, public_boundary, the four script tests, the adoption and grounding tests, npm test (6 passed)
• The gate binary against a mock coach API: 26 of 26 checks; the published 1.0.22 fails the 8 new ones on the same mock

Release binaries, wire diff, HTTP smoke and real-binary checks come next, before the tag.

🤖 Generated with Claude Code

---

^{Need help on this PR? Tag @codesmith-bot with what you need.}

[claude]

⚠️ Code review skipped — your organization has no extra usage available to pay for this review.

If your organization's extra usage balance is empty, an organization admin can add extra usage credits at claude.ai/admin-settings/usage. If its monthly spend limit was reached, an admin can raise it on the same page. If neither applies, contact Anthropic support.

Once extra usage is available, someone with write access to this repository can comment @claude review on this pull request to trigger a review.