refactor: Improve data repository lifecycle safety

[Himanth-reddy]

Original PR #11 Context

Source PR: Himanth-reddy#11

CodeRabbit walkthrough from PR #11:

Three independent repository class refactorings improve stream resolution logic clarity, add atomic lifecycle state management to prevent concurrent starts, and reorganize telemetry DataStore keys into a companion object.

Original Changes

Layer / File(s)	Summary
Stream resolution optimization app/src/main/kotlin/com/arflix/tv/data/repository/AddonRuntimeAggregator.kt	resolveMovieStreams and resolveEpisodeStreams now use early returns on empty input and direct STREMIO runtime resolution with .orEmpty(), removing intermediate mutable lists and conditional blocks.
Lifecycle thread-safety app/src/main/kotlin/com/arflix/tv/data/repository/CloudSyncCoordinator.kt	started state changed from volatile boolean to AtomicBoolean, with start() using atomic compareAndSet(false, true) for idempotent behavior and stop() resetting via set(false).
Telemetry key organization app/src/main/kotlin/com/arflix/tv/data/repository/PlaybackTelemetryRepository.kt	DataStore preference keys (startupSamplesKey, startupAvgMsKey, and telemetry counters) moved from instance-level properties into a private companion object.

Review Feedback Addressed

CodeRabbit flagged a remaining lifecycle race in CloudSyncCoordinator.kt: AtomicBoolean.compareAndSet serialized concurrent start() calls, but did not protect collectorJob and flushJob from a concurrent stop() call. That could allow start() to assign a live job after stop() had nulled the job fields, leaving coroutine work running while started == false.

CodeRabbit also suggested replacing the fully-qualified java.util.concurrent.atomic.AtomicBoolean usage with an import.

Additional Commit Added

Commit: c7d0944 fix: synchronize cloud sync coordinator lifecycle

This commit:

• Adds a shared lifecycleLock in CloudSyncCoordinator.
• Synchronizes start() and stop() so started, collectorJob, and flushJob are updated together.
• Synchronizes scheduleFlush() and skips scheduling if the coordinator has already stopped.
• Cancels/replaces flushJob under the same lifecycle lock.
• Imports AtomicBoolean directly and uses AtomicBoolean(false).

Validation

• ./gradlew assembleSideloadDebug --no-daemon