feat(logic): Implement Smart Recommendation Engine & Data Pipeline Upgrades

What does this MR do?

This MR implements the mandatory Recommendation Algorithm. Instead of using the generic TMDB recommendation endpoint, we built a custom "Content-Based Filtering" engine that analyzes the user's local history to suggest movies.

The Recommendation Algorithm (How it works)

The engine runs on a "Genre-Weighted" logic:

1. Data Harvesting: We fetch all movies from the local database that the user has Liked OR Rated.
2. Profiling: We calculate a "Score" for every genre found in these movies.
   • Like: +3 Points for that genre.
   • Rating: +[Star Rating] Points (e.g., 4.0 stars adds 4 points to the genre).
3. Targeting: We identify the user's Top 2 Genres (e.g., Sci-Fi and Action).
4. Discovery: We query the TMDB API using two strategies:
   • Strategy A (Comfort Zone): High-rated movies in the #1 Genre.
   • Strategy B (Discovery): Popular movies containing the #2 Genre.
5. Filtering: We merge the lists, shuffle them, and remove any movies the user has already seen.

Key Technical Changes

• Data Layer Upgrades:
  • Updated Movie models and Room Schema (Version 5) to store Genre IDs and User Ratings locally.
  • Updated MovieRepository to capture genres from the API.
  • Updated UserCollectionsRepository to implement the algorithm.
• Pipeline Fixes:
  • Fixed a critical bug in ActionBottomSheet where genre data was lost during the UI handoff. Now passing data via Bundle ensures saved movies are valid for the algorithm.
• UI Integration:
  • Updated HomeViewModel to prioritize Personalized Recommendations.
  • Added logic to show a "Rate movies to get recommendations" message if the profile is empty.

Important Note on User Behavior

The algorithm treats Ratings and Likes as separate signals.

• If a user Likes a movie, they get recommendations based on it.
• If a user Un-Likes a movie but keeps the Rating (e.g., 4 stars), the system still uses that movie for recommendations.
• Why? Because a high rating indicates preference, even if the movie isn't in the "Favorites" list. To stop recommendations based on a movie, the user must essentially "forget" it (delete from DB/Watchlist).

How to Test

1. Fresh Start: Clear app storage to remove old movies with missing genre data.
2. Seed Data: Search for a movie (e.g., "The Matrix"), open details, and Like it. (This saves the Sci-Fi genre).
3. Check Home: The recommendation row should now be populated with Sci-Fi movies.
4. Check Logs: Filter Logcat for RecEngine to see the scoring math in real-time.