Another question I have is whether it's possible to have a worker essentially drop a deployment run to allow another worker to pick it up. i.e. Worker A picks up a deployment in the queue, it then realises that something about it's infrastructure is balked, e.g. no internet. Instead of failing and retrying on the same worker a bunch of times unnecessarily, I want the worker to drop the flow and put it back to the queue. a la SQS message visibility timeouts.

You can use Automations for this. e.g. when flow state enters Crashed, change state to Scheduled if you have 2 distinct infrastructure methods you could instead trigger a different deployment, one that is the same flow code but deployed with a different infra block