Backend chat observability: add standardized request summary logs

[akgohain]

Motivation

• Provide consistent, single-line structured observability for chat endpoints so stalls/errors can be diagnosed quickly with minimal parsing.
• Ensure every chat request emits a compact summary containing request_id, endpoint, latency_ms, status, and error_type while preserving existing verbose behavior.
• Avoid logging sensitive request payloads or field context.

Description

• Add a module logger and request-timing/request-id helpers to server_api/main.py and emit a single-line JSON summary for /chat/query and /chat/helper/query in a finally block.
• Generate request_id from the incoming x-request-id header or a new uuid4, capture endpoint from req.url.path, compute latency_ms via time.perf_counter(), and record status (ok/error) and error_type when exceptions occur.
• Keep all existing endpoint logic and DB persistence unchanged; logging only emits metadata (no query, fieldContext, message content, or other payloads) to avoid sensitive data leakage.
• Add targeted tests tests/test_chat_logging_fields.py that capture logger output with self.assertLogs and assert presence and types of the standardized fields for a success path and an error path.

Testing

• Ran PYTHONPATH=. pytest -q tests/test_chat_logging_fields.py which passed (2 passed) verifying both success and error logging behaviors.
• Attempted uv run pytest -q tests/test_chat_logging_fields.py but it failed in this environment due to packaging/metadata for the editable pytorch_connectomics dependency (unrelated to the changes); the focused PYTHONPATH invocation was used to validate the new tests.
• No other automated tests were modified and existing endpoint behavior was preserved by the implementation.

---

Codex Task

[akgohain]

Consolidated into draft PR #165 (feat/codex-wave-integration). Closing this source draft to keep review in one place.