ci: sanitize PR head ref usage in calm-models publish workflow

[rocketstack-matt]

Description

Closes the OSPS-BR-01.01 / OSPS-BR-01.02 findings from the LFX Insights build & release review. release-calm-models-maven-publish.yml interpolated github.event.pull_request.head.ref directly into a run: script to extract the release version, and that derived value then flowed the same way into two further shell steps (tag + next-dev-iteration PR). A branch name containing shell metacharacters could inject commands into a job that has contents: write and Maven Central deploy credentials.

Fix: pass the ref and the derived version through env: in all three steps instead of template-interpolating them into the script body, and validate the extracted version against a strict MAJOR.MINOR.PATCH pattern before it's used anywhere — the job already gates on startsWith(head.ref, 'release-prep/calm-models-v'), so this catches any unexpected suffix.

Type of Change

• 🐛 Bug fix (non-breaking change which fixes an issue)
• ✨ New feature (non-breaking change which adds functionality)
• 💥 Breaking change (fix or feature that would cause existing functionality to not work as expected)
• 📚 Documentation update
• 🎨 Code style/formatting changes
• ♻️ Refactoring (no functional changes)
• ⚡ Performance improvements
• ✅ Test additions or updates
• 🔧 Chore (maintenance, dependencies, CI, etc.)

Affected Components

• CI/CD

Commit Message Format ✅

Used ci: (unscoped fix/feat/perf trigger releases per cli/.releaserc.json's fallback rule; ci is release-neutral).

Testing

• I have tested my changes locally
• I have added/updated unit tests
• All existing tests pass

This workflow only runs on a merged release-prep/calm-models-v* PR against main, so it can't be exercised in CI here. Verified: YAML parses correctly, and the shell logic (prefix-strip, semver validation, env: substitution) was traced by hand against the existing step outputs and Copilot CLI reviewed the diff with no findings.

Checklist

• My commits follow the conventional commit format
• I have updated documentation if necessary
• I have added tests for my changes (if applicable)
• My changes follow the project's coding standards

[Copilot]

Pull request overview

This PR hardens the release-calm-models-maven-publish.yml GitHub Actions workflow against shell-injection risks by avoiding direct interpolation of untrusted PR branch refs into run: script bodies and by validating the derived release version before use.

Changes:

• Pass github.event.pull_request.head.ref via env: and derive RELEASE_VERSION inside the shell, rather than embedding the ref directly into the script.
• Validate the extracted release version against a strict MAJOR.MINOR.PATCH regex before writing it to GITHUB_OUTPUT.
• Pass the derived RELEASE_VERSION via env: into later steps that tag the release and create the next-development-iteration PR.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.