release: 3.1.0

Bumps `_version.py` and `CITATION.cff` together, dates the changelog and adds the
compare link, then regenerates the gallery -- in that order, because
`examples/gallery/index.json` and `showcase.json` stamp `complexplorer_version` and a
regeneration before the bump leaves a stale manifest and a release commit that fails its
own suite. `release.yml` runs the artifact gate rather than the test suite, so nothing in
CI's release path would have caught that; only the test lanes do. This release is what
taught us the ordering, and the runbook now states it.

Both manifests changed by the version stamp alone: preset ids identical, no preset record
altered, no portrait churn.

Gate: 1019 tests pass, ruff clean, mkdocs --strict builds, openspec validate --specs
passes 20/20. The artifact gate was also run locally against the 3.1.0 build -- uv build,
check_distribution.py and twine check all pass, with 49 wheel entries against 3.0.0's 48,
the new `core/polyhedral` module.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

chore: restamp the gallery manifest for 3.1.0rc1

index.json carries complexplorer_version, and test_gallery asserts the committed
manifest reproduces byte for byte, so a version bump invalidates it. Only the
version line changed; the portrait PNGs are untouched, since they are reproducible
only best-effort and differ across machines.

This is the finding the rehearsal was for: CONTRIBUTING.md regenerates the gallery
(step 2) BEFORE bumping the version (step 3), which leaves the manifest stale and
the suite red on the release commit. The order needs reversing.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Merge pull request #10 from kuvychko/docs/openspec-baseline

Complexplorer 3.0.0

Merge pull request #8 from kuvychko/feature/new_cmaps

Version 2.0 release

Merge pull request #7 from kuvychko/feature/submodules_fix

Feature/submodules fix

Merge pull request #4 from kuvychko/feature/changelog

added changelog

edits to readme.md