Fix KeyError on API failures and pickle error in Python 3.13 (#9)

[parthashirolkar]

Summary

This PR fixes two critical bugs reported in issue #9:

Bug #1: KeyError when API times out or returns errors

• Problem: When the data.gov.in API returns errors (e.g., 504 Gateway Timeout), get_data() crashes with KeyError: 'total' because get_resource_info() returns an empty dict {} but the code tries to access ["total"] directly.
• Fix: Added proper error handling to check if resource_info is empty or doesn't contain "total" before accessing it. Now raises a clear ValueError with a helpful message.

Bug #2: Pickle error in Python 3.13 with multiprocessing

• Problem: The @retry decorator on make_request_with_retry() creates unpicklable state (an RLock), causing multiprocessing.pool.MaybeEncodingError: cannot pickle '_thread.RLock' object when pool.map() tries to serialize results from worker processes.
• Fix: Created _make_request_for_pool(), a non-decorated version of the request function specifically for multiprocessing workers. This avoids the pickle error while maintaining the retry logic for non-multiprocessing use cases.

Changes

• Added _make_request_for_pool() function for multiprocessing workers
• Modified get_api_records() to use the non-decorated function
• Added error handling in get_data() to check for missing "total" key
• Raises clear ValueError instead of crashing with KeyError

Testing

Both fixes have been tested:

• get_resource_info() now properly handles API failures without crashing
• get_data() works with multiprocessing in Python 3.13 without pickle errors
• All README examples work correctly

Fixes #9