Preserve all headers in response info

[jhamon]

Problem

Response headers were being filtered to exclude timing-dependent headers (x-envoy-upstream-service-time, date, x-request-id) to avoid test flakiness. However, these headers can be useful for debugging, monitoring, and understanding request behavior in production environments.

Additionally, the extract_response_info function was importing modules on every request and performing unnecessary checks, creating performance overhead for a function that runs on every API call.

Solution

Remove header filtering so all response headers are preserved in _response_info for REST, asyncio, and gRPC requests. This provides complete header information while maintaining correct equality comparisons (response dataclasses already exclude _response_info from equality checks).

Also optimize extract_response_info performance by moving imports to module level and removing unnecessary conditional checks.

Changes

Response Info Extraction (response_info.py)

• Removed filtering of timing-dependent headers (x-envoy-upstream-service-time, date, x-request-id)
• Optimized value conversion to check string type first (most common case)
• Updated documentation to reflect that all headers are now included

REST API Clients (api_client.py, asyncio_api_client.py)

• Moved extract_response_info import to module level to eliminate import overhead on every request
• Removed unnecessary if response_info: check since the function always returns a dict

Performance Improvements

• Eliminated import overhead by moving imports to module level
• Removed unnecessary conditional checks
• Optimized type checking order for header value conversion (check string type first)

These optimizations are especially beneficial for high-throughput applications making many API calls.

Usage Example

No changes required for users - the API remains the same:

from pinecone import Pinecone

pc = Pinecone(api_key="your-api-key")
index = pc.Index("my-index")

# All response headers are now available in _response_info
upsert_response = index.upsert(vectors=[...])
print(upsert_response._response_info["raw_headers"])
# Now includes all headers: date, x-envoy-upstream-service-time, 
# x-request-id, x-pinecone-request-lsn, etc.

query_response = index.query(vector=[...])
print(query_response._response_info["raw_headers"])
# Includes all headers from query response

Testing

• All existing unit tests pass (414+ tests)
• Integration tests verify response info functionality with all headers included
• LSN header extraction tests confirm all headers are preserved
• Tests access headers flexibly using .get("raw_headers", {}), so they continue to work with additional headers

Breaking Changes

None. This is a transparent improvement with no API changes. Response object equality comparisons are unaffected since _response_info already has compare=False in all response dataclasses (QueryResponse, UpsertResponse, UpdateResponse, FetchResponse, FetchByMetadataResponse).