Tweak staging and production frontend deployments so there is less downtime

[alalonde]

What would you like to be added?

Currently, when pushing to develop or master, there is this automated github actions flow:

• Undeploy frontend
• Delete frontend container
• Build frontend
• Deploy frontend

Building the frontend takes over 20m, which means at least 20m of downtime. Seems like we should build the new container before undeploying the old one.

Why is this needed?

We have a bunch of unnecessary downtime when updating staging and prod.

[alalonde]

This is fairly urgent, as a failed build now results in the previous instance being removed, and you have to fix the build to have ANYTHING deployed