docs(release): drop the last cross-refs to the removed alpha subcommand

[mocha06]

Motivation

The non-blocking item from the #556 approval: _apply_prepare's stamp note ended with a (see alpha) cross-reference to a subcommand that no longer exists. It was the one pointer a reader would follow to understand why stamp=False exists, and it led nowhere.

Sweeping for the same pattern turned up two more that the review did not catch, both in tests/test_release.py — comments crediting the ahead-of-main guard and the "only one flow cuts alphas" contract to alpha rather than alpha-pr. Those guards now live in the shared release_pr path.

Outcome

No comment, docstring, or doc in the release tooling names a subcommand that argparse does not register. The stamp note states its reason directly rather than pointing at a symbol, which is what made it rot in the first place.

prerelease_track's docstring deliberately keeps its alpha mentions: there the word names the PEP 440 track, not the subcommand.

Comment-only, so there is no CHANGELOG entry — no behavior changed, and the existing ## [Unreleased] bullet already describes the shipped contract. Adding one would only put noise in the next alpha's release body, since that body is ## [Unreleased].

[adriannoes]

The leftover from the #556 approval is gone: _apply_prepare's stamp=False note no longer points at a removed alpha subcommand, and the two test comments now credit release_pr / alpha-pr instead of the deleted one-step path. The rewrite states why Unreleased stays intact (beta promotion stamps the whole section) rather than naming a symbol that can rot again.

Checked clean against argparse (prepare, release-pr, alpha-pr, publish, verify), a leftover sweep of scripts/tests/RELEASE, and tests/test_release.py (49 passed). No behavior change, so skipping CHANGELOG is right. PEP 440 track alpha in prerelease_track correctly stays.