Feat: DevRank Active Pruning & Auto-Cleanup

Optimize the DevRank data lifecycle by actively pruning low-value users and enforcing hygiene before operations.

Problem:

1. Tracker Bloat: Updater.mjs currently keeps low-value users in tracker.json (updating their timestamp). This causes the index to grow indefinitely with users who have already been rejected.
2. Stale State: Manager.mjs runs update or spider operations potentially on "dirty" data (e.g., users that should have been pruned), leading to wasted processing.

Solution:

1. Pre-Run Cleanup: Modify Manager.mjs to automatically run Cleanup.run() before executing spider or update commands.
2. Active Pruning: Modify Updater.mjs to remove users from tracker.json immediately if they fail the contribution threshold, rather than just updating their timestamp.
3. Storage Support: Enhance Storage.mjs to support key removal in updateTracker (e.g., by accepting null or a specific removal flag).

Benefits:

• tracker.json stays small and focused on high-value + pending users.
• update runs never waste API calls on known low-value users (because they are pruned before the run or during the previous run).
• No need for a massive "Visited Users" database.

Tasks:

• Refactor Storage.mjs to handle deletions in updateTracker.
• Refactor Updater.mjs to request deletion for low-value users.
• Refactor Manager.mjs to inject Cleanup.run().