Add GitHub Actions CI workflow

[julien731]

Closes: #24

Summary

• Add GitHub Actions CI workflow with lint and test jobs
• Add Ruff to dev dependencies in requirements.txt
• CI triggers on PRs and pushes to develop and main branches

Approach

Two parallel jobs for faster feedback:

• Lint & Format: installs only Ruff, runs ruff check and ruff format --check
• Test & Coverage: installs backend + test dependencies (excluding ML packages which are mocked), runs pytest with coverage reporting and 80% threshold enforcement

ML dependencies (whisperx, torch) are excluded from CI since all tests mock the transcription pipeline (BR1).

Verification

• All 49 tests pass locally
• Coverage at 57% (below 80% threshold) — expected since integration/e2e tests are separate stories; CI will correctly enforce the threshold once coverage catches up
• Ruff check and format pass on the codebase
• Architect review passed with all critical/major issues addressed

[julien731]

QA Confidence Verdict: Issue #24 - CI Workflow

What Was Verified

#	Check	Status	Method
AC1	Workflow file exists at .github/workflows/ci.yml	PASS	Code inspection
AC2	Triggers on PRs to develop and main	PASS	Code inspection (also triggers on push per architect review)
AC3	Steps: checkout, Python 3.12, deps, ruff lint, ruff format, pytest+coverage	PASS	Code inspection
AC4	Workflow fails if lint/format checks fail	PASS	ruff check and ruff format --check exit non-zero on failure
AC5	Workflow fails if coverage below 80%	PASS	--cov-fail-under=80 flag present
AC6	Coverage report printed to stdout	PASS	--cov-report=term-missing flag present
AC7	Ruff added to dev dependencies	PASS	ruff>=0.9.0 in requirements.txt
BR4	Ruff is single tool for lint+format	PASS	Only ruff used, no flake8/black/isort
BR5	CI on PRs only (+ push to develop/main per review)	PASS	Triggers match

What Needs Human Eyes

Nothing significant. This is infrastructure config with no UI impact.

Risk Areas

Coverage is currently at 56.9%, below the 80% threshold. Running pytest --cov --cov-fail-under=80 locally fails. This means the CI pipeline will correctly fail on the current test suite. This is expected behavior (the workflow correctly enforces the threshold), but the team should be aware that CI will be red until test coverage is improved.

The CI test job hardcodes dependency versions rather than using requirements.txt directly. This creates a maintenance risk where CI deps and local deps could drift apart. However, this is intentional since the full requirements.txt includes ML packages (whisperx, torch) that are impractical in CI.

Suggested QA Focus

• Quick glance: Confirm the workflow runs successfully on this PR by checking the GitHub Actions tab
• Follow-up: Plan to increase test coverage to 80%+ so the CI gate passes on future PRs