Retry post_api and delete_api on InternalServerError

Dawid Rycerzrequested to merge
knightdave/gitlab-triage:fix-post-delete-retry-on-internal-server-error into master

What does this MR do and why?

Adds Errors::Network::InternalServerError to the retry exception_types for post_api and delete_api in RestAPINetwork.

query_api already retries on transient 500 errors, but post_api and delete_api only retried on Net::ReadTimeout. This meant that any transient InternalServerError from GitLab during a POST or DELETE would immediately fail the triage job, even though the same request would succeed on retry.

This was observed when running triage against a group with many epics — the first POST to create a note would occasionally return a 500, failing the entire job. Manually retrying the same request succeeded immediately.

The fix aligns post_api and delete_api with the existing query_api retry behavior.

Closes #396 (closed)

Merge request reports