Shouldn't we instead define a test status in which the application shows up in a different queue, then the owner can request approval to put it in the normal queue, then it can be approved? (That's T96157 if I understand the intent behind that ticket correctly.) It seems pointless to duplicate applications.

Or maybe having a separate test app would be useful for testing updates to an already approved app? (In that case, wouldn't it be better to just automatically add a different test key/mode for each app?)

In T60937#1391002, @Tgr wrote:

Shouldn't we instead define a test status in which the application shows up in a different queue, then the owner can request approval to put it in the normal queue, then it can be approved? (That's T96157 if I understand the intent behind that ticket correctly.) It seems pointless to duplicate applications.

This sounds like a sane way to go. If we have "Beta" (or "Test" or "Dev") as a stage that can be optionally set on the application form, and then allow on the management form to transition the app to "ready for review" (or something like that), I think that would address both this and T96157.

I think I also added a task about marking apps as "Personal user" or "Bot only"-- where the user who's making the request is the only intended consumer (so users can store the tokens instead of a username/password on their bot servers). That could be implemented in a similar way, so it might be good to start in this direction.

I think I also added a task about marking apps as "Personal user" or "Bot only"

Yes, T87395.

See T103587#2621901.

Change 316302 had a related patch set uploaded (by Gergő Tisza):
[WIP] Add development stage, refactor approval workflow

https://gerrit.wikimedia.org/r/316302