docs(contributing): draw pep.rst's release-trigger chain as a Mermaid graph

[JarryShaw]

Please follow the guide below

• Searched for similar pull requests
• Followed the coding style (make pylint, make mypy, make isort) -- N/A, docs only
• make test passes, and a test case covers the change
• Added a changelog entry under docs/source/changelog/ -- N/A — centralised in docs(changelog): shared 1.5.0 changelog — long-lived, merges last (#610, #616, #617, #618, #620) #657

What is the purpose of your pull request?

• docs — documentation only

---

Description

Part of #719 (docs/source/contributing/pep.rst slice), per the repo owner's ruling there: prefer a Mermaid graph over a paragraph/ASCII diagram where the subject is a flow.

Converted: the .. code-block:: text ASCII chain in "Release Plan" (push to main -> Vendor Update -> Create Release) to a .. mermaid:: flowchart, styled to match the three existing directives in workflows.rst/releasing.rst (flowchart TD, short node IDs, <br/> labels, thick ==> for workflow_run).

Factual fix found while verifying: the old diagram skipped a hop and misattributed a trigger. It said cron-vendor.yml triggers on push to main — but cron-vendor.yml:6-13's own comment says it deliberately carries no push trigger (to avoid double-running Unit Tests' gate job, #715). The real chain is push to main → Unit Tests → Vendor Update → Create Release, verified against cron-vendor.yml, create-release.yml, and unit-tests.yml directly.

Considered and rejected: the "Every publishing job… is gated on" paragraph (per-job if: rationale) and the "Delivery Sequence" wave ordering — both are "why" rationale a diagram can't carry, so left as prose. No other ASCII diagrams exist in the file. Checked unit-tests.yml:58-59 for the 3.10–3.14 matrix claim at pep.rst:552 — accurate, unchanged.

Cross-references :doc:workflows`` rather than duplicating its repository-wide graph.

[JarryShaw]

GOOD TO GO at 9dd2d62c8 — fable cross-review, first round clean. Different
model from the author (sonnet).

I re-derived its load-bearing claims rather than relaying them:

• cron-vendor.yml genuinely has no push trigger — only schedule and
  workflow_run on Unit Tests. The old prose was wrong, and it also skipped
  the Unit Tests hop entirely.
• The citation cron-vendor.yml:6-13 is exact: :5 is the cron entry, :6 the
  first comment line, :13 the last, :14 is workflow_run:.
• The Mermaid block matches workflows.rst:143-153 node-for-node, and
  sphinxcontrib.mermaid is loaded at docs/source/conf.py:81.
• releasing.rst:94-101 already described this chain correctly, so pep.rst was
  the odd one out rather than being fixed into disagreement.

No full Sphinx build was run on either side; the structural-subset argument plus
CI's docs leg is what stands behind the directive rendering.

Omitting the Saturday schedule edge is a scoping choice, not a defect — the
paragraph scopes itself to the ordinary-commit path and defers the full graph to
:doc:workflows, which already draws it.