[Index Management/ML] Fix broken retries on inference id creation

[sphilipse]

Summary

This fixes the inference endpoint creation API being called multiple times on error. The call will often time out because downloading/deploying the model takes longer than the Kibana request timeout limit. Setting the timeout limit higher would still be fragile, so ignoring the timeout error makes more sense.

This PR also contains a few small language fixes and variable renames for clarity.

[yuliacech]

Hey @sphilipse, I've tested locally and the timeout error seems to break the logic to save mappings with semantic text: on a new deployment, the request fails and it's not possible to refresh.

[yuliacech]

Adding a screen recording as well

Screen.Recording.2024-06-27.at.11.22.45.mov

[darnautov]

ML changes LGTM

[sphilipse]

Thanks @yuliacech, looks like the timeout error check function doesn't work for this specific API. Replaced that and this should work now :)

[kibana-ci]

💚 Build Succeeded

• Buildkite Build
• Commit: 4ad2282

Metrics [docs]

Async chunks

Total size of all lazy-loaded chunks that will be downloaded as the user navigates the app

id	before	after	diff
indexManagement	665.2KB	665.2KB	+21.0B

History

• 💛 Build #218484 was flaky ad2ef2f
• 💛 Build #218004 was flaky 3d92f96
• 💛 Build #217926 was flaky 5345b09
• 💔 Build #217856 failed 452e6d3

To update your PR or re-run it, just comment with:
@elasticmachine merge upstream

[yuliacech]

Thanks for fixing the timeout issue, @sphilipse! Re-tested locally, latest changes LGTM 👍