feat: stabilize v0.3.0 public API bundle

[StefanSteiner]

Summary

The v0.3.0 release rollup. This is the commit release-please will see and use to cut v0.3.0.

The four bundle PRs all landed under chore: prefixes to defer release-please from cutting an early version. This single feat: commit with a BREAKING CHANGE: footer aggregates the bundle and triggers the v0.3.0 release PR.

What's in v0.3.0

PR	What	Issue
#71	Flat public Error enum + ergonomic constructors workspace-wide	#70
#73	Transaction API consolidation — RAII guard recommended; raw trio #[doc(hidden)] deprecated	#69
#74	FromRow modernization — #[derive(FromRow)], RowAccessor, breaking trait signature, tuple impls deleted	#61, #62
#76	Follow-ups A/B/C — typed io::Error sources in process.rs, Error::InvalidOperation variant, structured SQLSTATE on Cancelled / Closed / Connection	—

Deferred to v0.3.x

#75 — flatten internal client::Error. Internal-only, zero external consumers, scope grew on second look (51 producers + 15 kinds vs. originally estimated ~6). v0.3.0 has all user-facing improvements without it.

The changes in this PR

Commit 1 (feat:): Two-line documentation refresh in hyperdb-api/src/lib.rs to (a) add hyperdb-api-derive to the companion-crates list (it was added in #74 but not mentioned at the crate level), and (b) fix a stale crate name (sea-query-hyper → sea-query-hyperdb). This is the minimum-viable code change to carry a feat: commit. The semantic content of the release is in the commit body's BREAKING CHANGE: footer, which release-please uses to generate the v0.3.0 entry in CHANGELOG.md.

Commit 2 (chore(ci):): Pre-release CI/CD audit found that hyperdb-api-derive (added in #74) was missing from release.yml's publish-in-dependency-order step. Without this fix, cargo publish -p hyperdb-api would fail at release time because hyperdb-api/Cargo.toml pins hyperdb-api-derive = "=0.X.Y" (strict equality) and the registry wouldn't have hyperdb-api-derive 0.3.0 yet. Also adds the derive crate to the publish dry-run job in ci.yml so the same class of bug is caught in CI next time, and refreshes a stale "7 workspace-member version rows" comment in release-please.yml.

CI/CD readiness audit (pre-merge)

Both adversarial reviewers (line-level + architect) ran against the v0.3.0 CI/CD config. Findings reconciled:

Severity	Finding	Status
CRITICAL	hyperdb-api-derive missing from release.yml publish list	✅ fixed in commit 2
MAJOR	Derive crate not in ci.yml publish-dry-run	✅ fixed in commit 2
NIT	Stale "7 workspace members" comment in release-please.yml	✅ fixed in commit 2
MAJOR (architect)	release-please-config.json extra-files doesn't list every workspace crate	Deferred to v0.3.x — every crate uses version.workspace = true so the workspace-root bump cascades automatically; the extra-files list only matters for crates with x-release-please-start-version markers (currently 3 crates, all listed). Filing follow-up.
MAJOR (architect)	npm-build-publish vs. release.yml race / no rollback path	Deferred — both workflows have idempotent re-run paths (already exists on guard for crates.io, will require similar for npm). Filing follow-up.
MAJOR (architect)	Hardcoded required-check list in two places — silent skip on rename	Deferred — would need a refactor against branches/main/protection API. Real but lower-stakes for the immediate v0.3.0 cycle. Filing follow-up.
MAJOR (architect)	Strict =X.Y.Z pinning + 45s settle window — could time out in registry-lag scenarios	Accepted as-is for v0.3.0; can add retry loop in v0.3.x.
MINOR (architect)	Lockfile-sync sentinel awk pattern fragile to future cargo formatting changes	Deferred — current cargo behavior is stable.
LOW (architect)	No cargo-semver-checks guard against accidental #[non_exhaustive] removal	v0.3.x improvement.

Verification

• cargo build --workspace --all-targets ✅
• cargo clippy --workspace --all-targets -- -D warnings ✅
• cargo test --workspace ✅ — 81 test groups pass
• cargo fmt --check ✅
• cargo doc --workspace --no-deps ✅ — 6 warnings (= post-Flatten Error type to canonical M-ERRORS shape (drop Client wrapper, Box<dyn> source, Option<ErrorKind>) #70 baseline)

What happens after merge

1. Release-please opens an aggregating chore: release main PR. The CHANGELOG.md entry on that PR will list:
   • The feat: from this commit (subject + BREAKING CHANGE footer)
   • All chore: commits from chore(api): flatten Error enum and add ergonomic constructors workspace-wide (#70) #71, chore(api): deprecate raw transaction trio; migrate MCP async paths to RAII (#69) #73, chore(api): modernize FromRow with #[derive] + RowAccessor (#61, #62) #74, chore(api): v0.3.0 follow-ups (A: typed I/O sources, B: InvalidOperation, C: structured SQLSTATE) #76 (release-please reclassifies them under the major-bump section because of the BREAKING CHANGE footer here)
2. Inspect the CHANGELOG.md draft. Tweak if needed; the per-PR MIGRATING-0.3.md sections are already comprehensive.
3. Merge the release PR. release-please tags v0.3.0, publishes crates (per the existing release.yml), and cuts a GitHub Release.
4. Update MIGRATING-0.3.md if any post-merge corrections are needed (likely none — it's been polished across 4 PRs).

Test plan

• All workspace gates green on macOS aarch64
• .release-please-manifest.json still at 0.2.3 pre-merge (verifying chore: strategy held — confirmed)
• Linux CI (will run on push)
• After merge: release-please opens a release PR titled chore: release main bumping to 0.3.0
• Inspect draft CHANGELOG.md — must list Flatten Error type to canonical M-ERRORS shape (drop Client wrapper, Box<dyn> source, Option<ErrorKind>) #70, Remove the &self begin_transaction/commit/rollback trio (RAII Transaction is the only safe path) #69, Add #[derive(FromRow)] and named-column access for cleaner struct mapping #61, FromRow redesign (breaking): typed RowAccessor, structured errors, enforced NULL semantics #62, chore(api): v0.3.0 follow-ups (A: typed I/O sources, B: InvalidOperation, C: structured SQLSTATE) #76 contributions
• Merge the release PR; v0.3.0 ships