Standardize procedure for using new apps (and new versions of current apps)

[owencking]

Summary

As a Chowda (super)user,
when an app (or new version of an app) is in the CLAMS app directory, but not yet in the list of available apps listed in the CLAMS Apps Chowda tab,
there is a standard procedure for deploying it in Metaflow (and/or doing whatever else is necessary) so that I can add a new row to the "CLAMS Apps" table (including the appropriate endpoint value),
so that I can use the (version of) the app I want.

Related

This is related to related to issue #196, about displaying the names for different apps.

Alternatives

No response

Additional context

Context and comment: Imagine there is one CLAMS app with ten versions, numbered v1 through v10. Suppose we know that v4 ran very fast but worked well only on a narrow set of assets. We might want to keep it around for that reason. We also might know that v8 worked great on a large set of assets, and we might want to use v8 again. Maybe v9 used a new model, but turned out to be problematic. Finally, suppose that we might want to use v10 or whatever is the newest version. In that scenario, I would want three different entries in the CLAMS app table, one with docker image :v4, and other with image :v8, and another with image :latest. It would be totally fine if these were just separate entries in the CLAMS apps table, as long as the version numbers where built into the names of the apps (and probably, I guess, also built into the names of the endpoints) -- so that we could tell them apart, and put them in the right pipelines.

[mrharpo]

@owencking I just happened to be working on this today!

Try out clamsproject/clams-deploy#1

I wrote some instructions in the README that should allow you to be able to generate the files. Let me know if it's clear enough, or if you have any issues going through the process.