v3: Refactor attempt creation to be worker requested

[ericallam]

• The "worker" side now is responsible for requesting a new attempt
• Currently this is only implemented on the "dev" side
• Moves the visibility queue outside of redis and into a graphile job, which we are now using to fail runs that are in progress and hit the visibility timeout (unless the run is pending)
• Better handle cancelling runs when dev reconnects to a different web server instance
• This should allow prod workers to retry without checkpointing and without going back into the queue

[changeset-bot[bot]]

🦋 Changeset detected

Latest commit: 2099d9162d36605dbb59baab1625bd4a32842d0b

The changes in this PR will be included in the next version bump.

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR