ci: publish to the MCP Registry after PyPI on each release

[JeanExtreme002]

Publishing to the MCP Registry used to be a manual mcp-publisher login github + publish after every PyPI release (3.0.2 was published that way). This adds it to publish.yml, so a GitHub release now ships to both.

The new mcp-registry job

It runs with needs: publish, because the registry validates the listing against PyPI: the README's mcp-name marker and the version that server.json pins.

1. Check the release against server.json. It fails if the tag isn't v + server.json's version, so a mismatched listing is never published. tests/mcp_server/test_registry.py already keeps server.json equal to __version__.
2. Wait for PyPI. PyPI's JSON API can lag the upload, so the job polls it (up to 10 minutes) until that version's description carries the marker.
3. Install mcp-publisher v1.8.1. It's pinned instead of latest, so a new CLI release can't change publishing unnoticed.
4. mcp-publisher login github-oidc + publish. The workflow's OIDC token proves the repo belongs to JeanExtreme002, which owns the io.github.JeanExtreme002/* namespace. No secret is stored. The job has id-token: write and contents: read only.

The workflow's display name is now Publish to PyPI and the MCP Registry. PyPI trusted publishing matches the file (publish.yml) and the pypi environment, neither of which changed.

Checked locally

• actionlint passes, shellcheck included.
• The tag check passes on v3.0.2 and fails on v3.0.3.
• The PyPI wait condition is true for 3.0.2 and false for 3.0.1, which has no marker.
• The v1.8.1 linux_amd64 asset downloads and lists github-oidc among its login methods.

The OIDC login itself can only run on a real release; 3.0.2 is already in the registry, so the next release is the first run.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.89%. Comparing base (31d31f1) to head (54fc973).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #104   +/-   ##
=======================================
  Coverage   89.89%   89.89%           
=======================================
  Files          41       41           
  Lines        3682     3682           
=======================================
  Hits         3310     3310           
  Misses        372      372           

Flag	Coverage Δ
Linux-py3.10	64.94% <ø> (ø)
Linux-py3.11	64.94% <ø> (ø)
Linux-py3.12	64.94% <ø> (ø)
Linux-py3.13	64.94% <ø> (ø)
Windows-py3.10	66.51% <ø> (ø)
Windows-py3.11	66.51% <ø> (ø)
Windows-py3.12	66.51% <ø> (ø)
Windows-py3.13	66.51% <ø> (ø)
macOS-py3.12	87.22% <ø> (ø)
mcp-Linux-py3.12	91.15% <ø> (ø)
mcp-Windows-py3.12	91.35% <ø> (ø)
mcp-macOS-py3.12	91.45% <ø> (ø)
speed-Linux-py3.12	66.54% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.