[Active Job Continuations](rails/rails#55127)
will be added in Rails 8.1. Queue adapters need to implement the
`stopping?` method to support them.

Implement the method and set it to true when a worker is stopping.

Also add Rails versions so we can test against Rails main. To prevent an
explosion of combinations, I've switched to only testing against minor
versions of Ruby.

… pruned

e.g. due to sleep/wake in a dev app

And point to the recommended way to do throttling.

Closes #560, #562

…t considered dead during a graceful shutdown

Remove 'duplicate' verbiage and use concurrency limits instead, simplify control flow

Fix race condition vulnerability by changing logic to enqueue

Add assertions when bulk enqueuing jobs with concurrency controls

Dispatch jobs in the order they were enqueued

Set ActiveJob successfully_enqueued for both enqueued/blocked and discarded jobs

Change concurrency 'at_limit' -> 'on_conflict'

Update discard logic to trigger an ActiveRecord rollback when attempting dispatch to prevent discarded job creation

Change default on_conflict concurrency option to old behaviour (blocking execution)

Add concurrent on_conflict documentation to README

Add test for discarding grouped concurrent jobs

Fix tests which expect raising enqueue errors

Add test to confirm scheduled jobs are also discarded

Co-authored-by: Philippe Tring <[email protected]>

Brought together #523 and #586, and took some parts from each.
Clean and enhance the README as well.

* fix: race condition with interruptibles

* fix: make attr_reader private

… and that it will be needed to look at it via the UI

Otherwise we'd get inconsistent behaviour with other adapters like
Sidekiq, where only jobs that are actually enqueued get assigned a
provider_job_id. Turns out, issuing a DELETE query for a record that got
inserted within a transaction works perfectly, even with bulk INSERT, so
let's just avoid creating those jobs altogether.

Add support for a `discard` option in concurrency controls