Report coverage on PRs with octocov

[gabrieldeespindula]

Follow-up to #284, which removed the broken qlty integration. This PR adds the coverage reporting that qlty used to provide, using octocov on top of the SimpleCov setup the project already has. It is an alternative to the couve-based flow proposed in #285/#286.

What it does

• On pull requests: octocov posts a single, auto-updated comment with the project total coverage and the delta vs master, plus a per-file breakdown of the files in the PR scope. The full SimpleCov HTML report (with uncovered lines, file by file) is uploaded as the coverage-html artifact of the run.
• On pushes to master: octocov stores the coverage report as a workflow artifact (artifact:// datastore), which becomes the baseline that future PRs are compared against.

How

• .octocov.yml: octocov reads SimpleCov's coverage/.resultset.json natively — no format conversion, no cc-test-reporter binary, no external service. Baseline storage and comparison use GitHub Actions artifacts with the built-in GITHUB_TOKEN (no secrets).
• Workflow triggers changed from push on every branch to pull_request + push on master — octocov needs the PR context to comment, and the master runs to record the baseline (same trigger change as Post coverage report as a PR comment via couve #285).
• Job permissions are scoped to the rspec job: pull-requests: write (comment) and actions: read (baseline artifact).
• The octocov step is skipped on PRs from forks, which get a read-only GITHUB_TOKEN and could not post the comment. Dependabot PRs are fine: the explicit permissions key is honored for them.

Notes

• The comment on this very PR will show no delta: the baseline only starts existing after this lands on master and the workflow runs there once. From the next PR on, the comment shows master X% → PR Y% (+/-).
• No coverage gate in this PR. When we want one, it is one line in .octocov.yml (e.g. coverage.acceptable: current >= prev), evaluated against the recorded baseline.

Tested

• Full suite locally: 291 examples, 0 failures; SimpleCov line coverage 99.5% (3598/3616).
• octocov v0.75.12 ran locally against the generated .resultset.json with this exact config: parsed it natively and reported 99.5%, exit 0.

🤖 Generated with Claude Code

[github-actions]

Code Metrics Report

Coverage	Test Execution Time
99.5%	12s

---

Reported by octocov