Update message when WASM returns an error and allow user to restart the server

[sejas]

Related issues

• Fixes https://github.com/Automattic/dotcom-forge/issues/10037

Proposed Changes

• Reject promise when the server dies when starting a server, so it can be started again by the user.
• Change the message displayed to the user when we get the "unreachable" WASM instruction executed

Before this PR, when a site crashed, it got unresponsive, and users couldn't re-try starting it again.
The reason is that the server child process failed and didn't resolve or rejected a promise when stopping the site, which let the main thread waiting for messages. This was fixed on : a4afa34

Testing Instructions

• Download a backup with a plugin that fails already activated. There is an example in the issue.
• Run npm start
• Create a blank site
• Go to the import tab
• Drag and drop the backup file
• Wait until the import succeeds
• Observe the site tries starting
• Observe a popup error saying : "[...] Please try disabling plugins and themes that might be causing the issue."
• Close the popup, wait a couple of seconds
• Click on start again
• Observe the same popup

Before

After

Pre-merge Checklist

• Have you checked for TypeScript, React or other console errors?

[sejas]

We could translate the error message here with __, but we probably prefer to log, and capture in Sentry all the messages in English.

[wojtekn]

Should we move the Sentry call here, too, so we would log it before we rethrow the exception?

[sejas]

@wojtekn , thanks for the suggestion. It makes sense. I've moved it here:

9c301a2

[wojtekn]

Looks good. In the future, we could handle other exceptions there with custom messages, or maybe remove the condition and use unified message for all case.

[sejas]

Agreed. I thought about overwriting all the errors, but I prefer to make the distinction in case we found any other error now or in the future after a dependency update.