Add caching layer for repeated PURL metadata queries

Problem : -

Currently, PurlDB processes repeated queries for the same PURL without any caching mechanism. This leads to redundant database/API calls, especially during large-scale operations such as SBOM imports or repeated metadata lookups.

Proposed Solution :-

Introduce a caching layer to store frequently accessed PURL metadata. This could include:

- In-memory caching (e.g., LRU cache)
- Optional Redis-based caching for scalability
- Cache invalidation strategy when data updates

The caching should be applied to:
- metadata queries
- version lookups
- URL retrieval functions

Benefits :-

- Improved performance for repeated queries
- Reduced database/API load
- Faster SBOM processing
- Better scalability for large datasets

Additional Context :-

Similar issues around repeated metadata access and inefficiencies have been observed in related tools interacting with PurlDB. Adding caching would improve overall system responsiveness and usability.