feat(helm): update chart to use Infrahub wrapper

[fatih-acar]

This currently depends on an upstream chart change.

I've pushed a PR upstream here: PrefectHQ/prefect-helm#449
Meanwhile, the new chart is available on our registry for testing/dev.

[codspeed-hq]

CodSpeed Performance Report

Merging #5740 will not alter performance

_{Comparing fac-update-helm-prefect-2 (7e80dd4) with develop (c93eed0)}

Summary

✅ 11 untouched benchmarks

[ogenstad]

One thought around this is if we should have an entrypoint that's static, either a shell script or some "infrahub prefect-server start" that would remain the same across versions even if we wanted to tweak some parameters. Basically something to avoid having do do what I did with the environment variables in the migration tests. And also so that we don't have to change the command everywhere if anything is needed to be updated in the future.

[fatih-acar]

One thought around this is if we should have an entrypoint that's static, either a shell script or some "infrahub prefect-server start" that would remain the same across versions even if we wanted to tweak some parameters. Basically something to avoid having do do what I did with the environment variables in the migration tests. And also so that we don't have to change the command everywhere if anything is needed to be updated in the future.

Using a shell script for the entrypoint would avoid having these upstream chart changes, if we reuse the same path as the original prefect-server image for the entrypoint: /opt/prefect/entrypoint.sh

We can probably go with that instead, I don't have a strong opinion.

[ogenstad]

I'll let you decide as it's mostly related to deployment :)

Though I think it would make sense to have our own entrypoint script instead of using the one from prefect as is. Mainly to avoid issues such as installation order as I think we'd have to overwrite Prefects entrypoint after the installation of the packages.

[fatih-acar]

Alright, the upstream Prefect Helm changes have been merged, so I'll go with best of both: having a custom entrypoint and use that same entrypoint in our Helm chart configuration. Will push changes for that.