Replaces the separate create_app/update_app/deploy_app calls with a single create_and_update_app call, making the code more DRY.

No API change or behavior change... other than that we're hoping for a side-effect from create_and_update_app's 0.10s sleeps (which had not been present in deploy_one_click_app):

We intermittently hit CapRover's "No NodeId was found" error (see caprover/caprover#930) between create_app and the first update_app, specifically when the app uses persistent data. My read from ServiceManager.updateAppDefinition is:

1. When create_app registers a new app, CapRover appears to spin up a placeholder service for it.
2. create_app -> _wait_until_app_ready polls on isAppBuilding but not on Swarm's task scheduling, so it can return before the new task has actually been assigned a NodeID.
3. When update_app fires in that gap, the task exists (so CapRover doesn't return "service not found") but NodeID is still empty, hence the error.

The 0.10s sleep isn't a guaranteed fix for this gap, but it happens to sit in the right place, so worth a try. Even if it doesn't fix "No NodeId was found", this PR is still a simple DRY improvement.