Before taking CloudLaunch out of beta...

[afgane]

Here is a summary of outstanding issues to address before we can move the current instance of CloudLaunch from https://beta.launch.usegalaxy.org/ to https://launch.usegalaxy.org/:

• Associate credentials with a deployment. If not done but multiple credentials/accounts for a given cloud are mixed by a user, health checks and other controls don't work, leading to a situation like this:
  

• Deployments stuck in pending state with no ability to control them. Unknown reason yet but some tasks are just stuck and a user has no way to archive or delete them:
  

• Checking health on a deleted/unknown deployment errors. When a health check is run on a deployment that's been deleted by means other than via CloudLaunch, an error is reported. Instead, check is the requested instance is missing and report it as MISSING. Current behavior:
  

• Expire celery tasks. We set a max runtime on tasks but need to also ensure that pending tasks eventually expire so set that to something like 5 minutes. Should be a doable by simply adding expires argument to the task invocation.

• Show more details about instance types. Currently, for AWS in particular, there are dozens of instance types available and all that's displayed in the dropdown is their name (e.g., m5.large) so at the very minimum add some specs describing the characteristics of the instance (e.g., num CPUs, amount of RAM, etc.):
  

• Allow order of appliances to be set. Currently, appliances are ordered alphabetically based on their name but to make more frequently used appliances more visible and easily accessible, allow explicit instance ordering.

• Fix default region selection on AWS. In particular c5 instance type in us-east-1 is not available in the e zone, yet our logic defaults there causing a launch to fail. Do something better default about zone selection.