Add retrying to the upload

[robertbrignull]

This should make the upload retry a few times and with a small amount of delay between attempts. Hopefully this will help get over very minor issues, though if there's a larger outage of github lasting more than a few seconds then it won't be able to help.

Merge / deployment checklist

• Run test builds as necessary. Can be on this repository or elsewhere as needed in order to test the change - please include links to tests in other repos!
  • I created branches to simulate various failure methods.
    • upload_retry where everything is normal
      • CodeQL using init/analyze actions - https://github.com/Anthophila/test-electron/actions/runs/92974577
      • 3rd party tool using upload action - https://github.com/Anthophila/test-electron/actions/runs/92974579
    • upload_retry_error where all attempts fail
      • In the logs you can see the final error message and that it retries with the expected delays.
      • CodeQL using init/analyze actions - https://github.com/Anthophila/test-electron/actions/runs/92961566
      • 3rd party tool using upload action - https://github.com/Anthophila/test-electron/actions/runs/92961572
    • upload_retry_partial_error where the first attempt fails
      • CodeQL using init/analyze actions - https://github.com/Anthophila/test-electron/actions/runs/92971009
      • 3rd party tool using upload action - https://github.com/Anthophila/test-electron/actions/runs/92971012
• Confirm this change is backwards compatible with existing workflows.
• Confirm the readme has been updated if necessary.

[anaarmas]

I see we're re-trying not only on 5xx responses but on all. Additionally, the behaviour we had of not setting the job as failed when we get a 500 is changed too. Can you provide more of an insight as to why you've decided to go this way?

[anaarmas]

will be retry -> will retry

[robertbrignull]

I see we're re-trying not only on 5xx responses but on all. Additionally, the behaviour we had of not setting the job as failed when we get a 500 is changed too. Can you provide more of an insight as to why you've decided to go this way?

Yes, thanks for spotting that. I completely forgot to mention that I made that change as part of this. I can easily back that change out if you'd prefer not to make it at this time.

I believe that the will of @jhutchings1 and @joshhale was to now fail the action if there's a 500 when uploading.

[robertbrignull]

In fact, I will back this change out and we can make it at another time.

[robertbrignull]

Changed to only retry on 5xx errors, and I've changed the error message to include the status code.

I've run some tests again so I'm happy with the testing of this.

[anaarmas]

why do we need this continue here?

[robertbrignull]

To get to the next iteration of the loop. Technically it's not necessary as this is the last statement of the loop in this branch, but it doesn't hurt as far as I can see.