Trim RELEASE.md to only repo-specific content

[jnasbyupgrade]

Summary

The org has consolidated the shared PGXN-extension release process into a
canonical ../ai/RELEASE.md (Postgres-Extensions/ai#8, not yet merged).
Per that doc's policy, a repo should only keep a local RELEASE.md for a
genuine repo-specific reason — not as a copy of steps the shared doc now
covers.

This repo's RELEASE.md was almost entirely a copy of the generic
pgxntool release steps. This PR trims it down to the one thing that's
actually specific to count_nulls: the distribution-version-vs-extension-version
distinction, including the concrete 0.9.7 historical example (a
distribution-only PGXN release where the extension version stayed at
0.9.6) — HISTORY.md's own 0.9.7 entry references this note directly,
so it's kept rather than deleted outright.

Depends on Postgres-Extensions/ai#8 landing first, since the new intro
links to ../ai/RELEASE.md. Opened as a draft for that reason.

Test plan

• Verified the 0.9.7 distribution-only claim against META.in.json history and HISTORY.md
• Confirmed HISTORY.md's reference to "RELEASE.md's note on distribution vs. extension versions" still resolves after the trim
• N/A — docs-only change, no code/build impact

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 809915e3-965f-473b-b971-a6673172e72a

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

Claude is reviewing this PR

• Gather context (PR diff, RELEASE.md history, HISTORY.md)
• Understand the request (code review via /code-review custom instruction)
• Review the trimmed RELEASE.md content
• Verify the 0.9.7 claim and cross-references
• Post review feedback

[github-actions]

Claude finished @jnasbyupgrade's task in 1m 26s —— View job

---

Claude finished the review

• Gather context (PR diff, RELEASE.md history, HISTORY.md)
• Understand the request (code review via /code-review custom instruction)
• Review the trimmed RELEASE.md content
• Verify the 0.9.7 claim and cross-references
• Post review feedback

Review

Clean, well-scoped trim. Only RELEASE.md is touched (+15/-135); nothing else in the repo changes.

Verified:

• HISTORY.md's 0.9.7 entry (line 22: "see RELEASE.md's note on distribution vs. extension versions") still resolves correctly — the trimmed RELEASE.md keeps the "Distribution vs. extension version" section (RELEASE.md:8-19) verbatim in substance.
• The kept content genuinely is repo-specific (the 0.9.7 distribution-only / 0.9.6 extension-version split) and isn't duplicated in generic pgxntool docs.
• The second commit (558b2f9) is a pure wording cleanup of the first ("count_nulls is one of the repos that tracks..." → "count_nulls tracks...") — no substantive change, good tightening.
• Everything removed (safety-check steps, pre-release checklist, tag/dist mechanics, stable pseudo-version dance) is generic pgxntool process, consistent with the PR's stated goal of deferring to ../ai/RELEASE.md.

Note (not a blocker, already disclosed by the author): the new intro at RELEASE.md:3 links to ../ai/RELEASE.md, which doesn't exist yet since Postgres-Extensions/ai#8 hasn't merged. That's why this PR is correctly opened as a draft. Once ai#8 lands, worth a quick sanity check that the section headings/anchors this trimmed doc implicitly relies on (e.g. versioning, make tag/make dist, the stable dance) actually exist there under those names — but that's a follow-up check against the other repo, not something blocking this PR.

No bugs, no factual issues found. I couldn't independently verify the 0.9.7/0.9.6 history against git tag/META.in.json history in this sandbox (shallow clone, no tags fetched), but the claim is unchanged from the pre-existing content and matches HISTORY.md's own 0.9.7 entry, which predates this PR.

[github-actions]

Claude finished @jnasbyupgrade's task in 1m 19s —— View job

---

Claude finished the review

• Gather context (PR diff, RELEASE.md history, HISTORY.md)
• Understand the request (code review via /code-review custom instruction)
• Review the trimmed RELEASE.md content, including latest commit c9d3e8e
• Verify the 0.9.7 claim and cross-references
• Post review feedback

Review

Still a clean, well-scoped trim — only RELEASE.md touched (+15/-137). Since the last review pass covered commits 881ea87/558b2f9, one new commit landed: c9d3e8e.

What c9d3e8e changes: it reframes the section heading from "Distribution vs. extension version" to "Distribution vs. extension version actually diverged once", and rewrites the body to explicitly attribute the generic distribution-vs-extension-version concept to ../ai/RELEASE.md step 3, leaving only the count_nulls-specific fact (that the split has actually been exercised, via the 0.9.7/0.9.6 example) as local content. This is a good tightening — it removes the last bit of duplicated generic explanation and makes the "why is this file not just deleted" rationale sharper.

Verified:

• HISTORY.md:22's reference ("see RELEASE.md's note on distribution vs. extension versions") still resolves — the note is still present under the new heading, just reworded to be tighter.
• The 0.9.7/0.9.6 claim is byte-for-byte unchanged across all three commits, so the earlier verification against HISTORY.md's own 0.9.7 entry still holds.
• No other files changed; nothing else in the repo is affected by this rewording.

Note (not a blocker): c9d3e8e's new wording — "../ai/RELEASE.md step 3 covers..." — hardcodes a step number from a doc that doesn't exist yet (Postgres-Extensions/ai#8 is still unmerged). This is a slightly tighter coupling than the previous wording (which didn't cite a step number), so it's a little more likely to go stale if that doc's step ordering shifts before/after merge. Given the PR is correctly still a draft pending ai#8, this is just something worth a quick glance at merge time — confirm step 3 in the landed ai/RELEASE.md is in fact the version-numbers step before undrafting.

No bugs or factual issues found.