fix: Deleting an asset is very slow

[shreyanshsingh1234-dev]

Summary
Optimized the performance of resource deletion (e.g., datasets) by refactoring many-to-many orphan deletion triggers and adding missing database indexes.

The Problem
Previously, deleting a resource triggered a full-table scan on all related tables (like keyword, publication, etc.) to check for orphans. This caused significant latency (measured in seconds) that would have become a major bottleneck as the database scaled.

The Solution

1. Targeted Triggers: Moved orphan deletion logic from the main resource tables to the many-to-many link tables. Triggers now only verify the specific items that were just unlinked.
2. Database Indexing: Added index=True to the linked_identifier column in all dynamically generated many-to-many link tables.
3. Efficiency: Reduced the orphan check complexity from O(N) (whole table scan) to O(1) (indexed lookup).