Improvement: Optimize LLM API calls by caching identical requests

[SHAURYASANYAL3]

If the prompt and context are exactly the same (e.g. after a dry run), cache the response locally.

[SakethSumanBathini]

@sreerevanth Can you assign me this issue