release: bump 0.1.3 -> 0.1.4

[kingchenc]

Two commits: the documentation that described how r-universe builds this package, and the version bump that ships the correction.

Why there is anything to correct

bindings/r/configure downloads the C ABI from the GitHub release named by DESCRIPTION: Version. The registry tracked this repository's default branch, where a version bump lands before the release it names has been built -- so every build of a bump commit resolved a 404 and the deploy status went red.

Over the last fourteen commits on main, the three release bumps are the only red ones. The middle of them was spared because the builder does not pick up every commit, not because it would have worked.

Nor is it a race a longer retry closes. 0.1.1 lost its GitHub release to a failed wasm publish, so main carried a version with no release at all and the registry stayed red until 0.1.2 replaced it.

The registry entry now carries "branch": "*release" (wickra-lib.r-universe.dev#1). release.yml drafts a release, attaches every asset, and publishes it only after all eight publish jobs have succeeded -- so a tag becomes visible to the registry with its assets already attached, and a release that never completes leaves the previous one in service instead of breaking the current one.

What this repository changes

Nothing that affects what the package installs. The registry lives in another repository; what changes here is that two files stated the old arrangement as a standing fact:

• CONTRIBUTING.md said r-universe builds bindings/r from the default branch. It now says which release it builds, carries the branch field in the registry snippet, and gives the reason.
• scripts/check_r_abi_skew.py justified its second, non-blocking check with "r-universe stays red until the next one". That consequence no longer exists. The check is unchanged and still warns -- main being ahead of the last release is a release-readiness signal; it simply no longer breaks anything while it waits.

Bump

bump_version.py reports 21 declarations rewritten, scripts/check_version_sync.py agrees on all 21. The only remaining 0.1.3 in the tree is a third-party crate that happens to be at 0.1.34.

Verification

• cargo fmt --all -- no changes
• cargo test --workspace --all-features -- 22 suites pass
• cargo clippy --workspace --all-targets --all-features -- -D warnings -- clean
• scripts/check_r_abi_skew.py -- wrapper links the released ABI unchanged

[codspeed-hq]

Merging this PR will not alter performance

✅ 2 untouched benchmarks

---

_{Comparing release/0.1.4 (fd6ae43) with main (a89b3d9)}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!