setup-nextstrain-cli: allow nextstrain update to continue-on-error

[joverlee521]

Description of proposed changes

I missed in #48 that the command nextstrain update will exit with error for aws-batch and ambient runtimes. Instead of maintaining a list of runtime to run the update command, just allow it to continue-on-error.

I found this error in separate work to update forecasts-ncov to use the pathogen-repo-build workflow with the aws-batch runtime.

Checklist

• Checks pass

[joverlee521]

Tested this in forecasts-ncov workflow

Green checkmarks throughout the run even though the nextstrain update command failed. The Annotations section of the summary page does note the error, but the link to the log line is incorrect because of actions/runner#1901

[corneliusroemer]

I find that error in the annotations summary page confusing. Would it be possible to do something about that? Like add a note to the annotation: "this is an expected failure" or the like?
See: https://bedfordlab.slack.com/archives/C0K3GS3J8/p1695495002980659

[victorlin]

I can't think of an easy way to hide or rewrite the error. I think the proper solution is to remove this nextstrain update step (it's only in place because the conda runtime does not use the latest version during nextstrain setup, as mentioned in #48).

[corneliusroemer]

@victorlin wasn't the update step added to address the conda issue? We could fix the root cause there, or alternatively: make that setup step conditional on conda runtime being used?

@joverlee521 wrote:

Instead of maintaining a list of runtime to run the update command, just allow it to continue-on-error.

In my view the confusing error message is worth maintaining a list of 1 that will need to almost never be updated 🙃

[tsibley]

wasn't the update step added to address the conda issue? We could fix the root cause there

The root cause was already fixed by nextstrain/cli#312. I believe the nextstrain update is now extraneous.

[victorlin]

@tsibley oh I missed that, thanks!