Fix/ resolve race condition in /generate-blog endpoint

[Unnati1007]

Description

This PR resolves a race condition in backend/main.py within the /generate-blog endpoint.

Previously, the code checked the database for an existing successful record using find_one(). However, because there was no concurrency locking mechanism, if multiple identical requests hit the server at the exact same time, all of them passed the find_one() check simultaneously. This resulted in multiple expensive parallel calls to the AI provider and identical duplicate posts being pushed to the publishing platforms simultaneously.

Fix #184

Implemented a distributed atomic lock using MongoDB's unique _id indexing.
Before processing a blog, the endpoint attempts to insert a record into a dedicated locks collection using a unique ID derived from the problem title, author, and user email. Only one request can successfully insert this lock. All other concurrent requests instantly receive a DuplicateKeyError and are cleanly rejected with a 400 Bad Request indicating the blog has already been published.

Steps to Reproduce

1. Start the FastAPI server locally (uvicorn main:app --port 10000).
2. Temporarily bypass authentication (for testing purposes) or pass a valid JWT token.
3. Run a Python script utilizing the threading and requests library to send 5 concurrent HTTP POST requests to /generate-blog simultaneously with identical JSON payloads.
4. Observe that exactly 1 request acquires the lock and begins processing, while the other 4 requests are immediately rejected by the lock.

[Unnati1007]

@vanshaggarwal27 please check the pr and merge it