Sync pgxntool to 2.3.0

[jnasbyupgrade]

Summary

• git subtree pull pgxntool from https://github.com/Postgres-Extensions/pgxntool.git release tag, updating the embedded copy from 2.2.0 to 2.3.0. Conflicts resolved by taking upstream's version for every file under pgxntool/ (it's vendored, no local patches to preserve) -- same convention as Sync pgxntool to 2.2.0 #21.
• Ran pgxntool/update-setup-files.sh to reconcile setup-copied files against the new version: pruned pgxntool/.claude/ and pgxntool/.github/ (pgxntool's own dev-only tooling, not meant to be embedded in consumer projects). No other setup files (.gitignore, test/deps.sql) needed changes this time.
• Bottom of a 4-branch stack restructuring Add test/install foundation for update+upgrade testing (fresh/update/existing) #22/CI: binary pg_upgrade testing, docs-only gate, single-source PG matrix #23/Add test/build: extension-metadata tests + raw SQL syntax checks #26 on top of this sync (coordinator is handling gh stack link separately). Pushed directly to this repo (not just the fork) so the next branch in the stack can reference it as a base without a fork-to-fork PR, same pattern already used for advanced-testing/foundation.

Why this matters for this repo specifically

Two real bugs are fixed in 2.3.0 that are directly relevant to work stacked on top of this branch:

• Issue #83 ("Fix installcheck running before install"): adds an explicit installcheck: install dependency edge in base.mk. This is the same root cause independently found and filed as Postgres-Extensions/pgxntool-test#62 -- check-stale-expected: installcheck's early edge could resolve installcheck before install ever ran on a genuinely fresh system. Confirmed present: grep -n "installcheck: install" pgxntool/base.mk -> line 326.
• Issue #48 (DATA wildcard, sql/*--*--*.sql -> sql/*--*.sql): fixes historical single-version install files silently dropping out of make install's DATA list once they're no longer the current version. Doesn't bite test_factory today (only one version, 0.5.0, has ever shipped), but would silently break TEST_LOAD_SOURCE=update testing (added by advanced-testing/foundation, stacked above this branch) the instant a second version ships -- make install would quietly stop installing the old version's SQL file, and an update test would have nothing to CREATE EXTENSION VERSION 'old' from. Worth calling out explicitly since it's not obvious from the pgxntool changelog alone that this repo has any exposure to it yet.

Verification

• make builds cleanly.
• make test passes (3/3) on PG17, from a machine where the extension had never been installed before.
• make verify-results passes cleanly.
• make pgxntool-version reports 2.3.0.

Test plan

• make
• make test on a machine with no prior install
• make verify-results
• CI (this PR)

[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: 3f0bf495-b261-4bb8-88ca-60c3059f1dc2

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.}

[jnasbyupgrade]

Superseded by #32 -- recreated as an upstream-branch PR so it can be part of a formal GitHub stack (gh stack link), which refuses PRs headed from a fork. Same content/commits, same CI results.