fix(github): ignore 404 when collecting unfinished pull requests (#9140)

[Thundercloud12]

Summary

When collecting unfinished pull request details (CollectUnfinishedDetails in pr_collector.go), DevLake iterates over previously collected open PRs (state != 'closed' OR merged = false) and queries:

GET /repos/{owner}/{repo}/pulls/{number}

If a PR has since been deleted, scrubbed (e.g., credential leak removal), or made inaccessible on GitHub, GitHub returns HTTP 404 Not Found.

Without an AfterResponse hook, ApiAsyncClient retries the endpoint 3 times and then terminates the entire subtask with:

Retry exceeded 3 times...

Consequently, all subsequent syncs fail on the same unavailable PR, causing data collection for newer PRs to halt and repository dashboards to become stale.

This PR adds AfterResponse: ignoreHTTPStatus404 to CollectUnfinishedDetails in pr_collector.go. The hook returns api.ErrIgnoreAndContinue when GitHub responds with HTTP 404, allowing the collector to gracefully skip unavailable PRs and continue collecting the remaining pull requests.

Does this close any open issues?

Closes #9140

Screenshots

N/A

Other Information

• Follows the existing convention used across DevLake collectors (e.g., cicd_job_collector.go).
• Added unit tests in plugins/github/tasks/shared_test.go covering ignoreHTTPStatus404 and ignoreHTTPStatus422.

[Thundercloud12]

@klesh hey so this is the one issue that you have assigned to me, applied the approach we discussed earlier in that thread, hope this helps!!

[Thundercloud12]

hey @klesh sorry for asking it here but the link to the slack discussion is expired i think can you guide me towards a working one

[klesh]

Good work. Thanks for your contribution.

[klesh]

hey @klesh sorry for asking it here but the link to the slack discussion is expired i think can you guide me towards a working one

Try the following one please:

https://join.slack.com/t/devlake-io/shared_invite/zt-47ktd6ws4-BdpD0M4W4eaJl8tGwH7Clw