CI: run lint, detekt, and ktlint on PRs

[patrickrb]

Follow-up from #27 / PR #58.

CI currently runs testDebugUnitTest + assembleRelease. No static
analysis — Android lint warnings are silent, Kotlin code-style isn't
enforced, and obvious correctness issues (NPE risks, unused params)
sneak through.

What to do

• Android lint: add ./gradlew lintDebug to the workflow. There's
  already a lintVitalRelease running as part of assembleRelease —
  the broader lintDebug is what's missing.
• detekt: add the detekt Gradle plugin + a baseline config focused
  on correctness rules (potential-bugs, exceptions, coroutines).
  Generate a baseline file initially so existing warnings don't gate
  the PR — new code is what we want to police.
• ktlint: add the ktlint Gradle plugin with the standard Kotlin
  style rules.

Phase the rollout: land each tool in warn-only mode first (CI runs
the task but doesn't fail on warnings), let the team see a few PRs'
worth of output, then flip to blocking. Don't introduce three new
blockers in one PR.

Why this matters

Catching issues before review is cheaper than the review pass. Style
enforcement also removes a class of review nit-picking.