Handle GitHub API exceptions in feature-links task

[love12yadav]

Summary

Fixes the feature-links cron/task pipeline so that a single failed GitHub API call (most commonly caused by rate limiting) no longer interrupts an entire batch of link updates.

Addresses #6632

Problem

internals/link_helpers.py's Link.parse() only handles a fixed set of exception types (requests.RequestException, HTTPError, ValueError, APIError).

When GitHub rate-limits requests (403/429), the underlying client may raise an exception outside this handled set. The exception then propagates through _index_feature_links_by_ids() in internals/feature_links.py, which previously had no exception handling.

This caused the /tasks/update-feature-links handler to return a 500 response, triggering Cloud Tasks retries on the same batch of links. The task would repeatedly fail instead of degrading gracefully.

Changes
internals/feature_links.py
Wrapped the per-link Link(...).parse() call inside _index_feature_links_by_ids() with try/except Exception so unexpected failures no longer terminate the entire batch.
Added _http_status_code_from_exception() and _normalise_http_status_code() helpers to extract HTTP status codes from common exception formats (.status_code, .status, .code, and nested .response objects).
Detect 403/429 rate-limit responses from:
raised exceptions, and
Link.parse() results (is_error / http_error_code).
Stop processing gracefully when rate limiting is detected, logging a warning with the number of remaining unprocessed links. Remaining links will be retried during the next scheduled run.
Handle non-rate-limit failures on individual links by recording the error on the corresponding FeatureLinks entity (is_error / http_error_code) and continuing processing of remaining links.

Fixes #6632