Allow apps to override / add to interruption decision

[djmortonShopify]

Right now, a job is interrupted if it runs longer than the configured JobIteration.max_job_runtime or if the relevant interruption adapter indicates the job should (due to a shutdown being invoked, for example).

Some apps may need finer grained control over this, or may have additional signals that should be inspected to determine when a job should stop iterating and reenqueue itself.

The way that I've seen this done typically is to override the private job_should_exit? method which is a little cludgy. Firstly, because it requires overriding a private method, but secondly, because if multiple concerns (say two different mix-ins) need to add behaviour, each override needs to make sure it calls 'super' to ensure that all relevant versions of job_should_exit? are called.

It might be a good idea to add some form of public hook that could be configured by the application to allow application authors to add functionality to the job_should_exit? check as necessary in their application.

One possible approach may be to add a chain of responsibility, where procs that determine whether the job should exit can be added and then each could be called in sequence whenever job_should_exit? is called, to determine if any of them indicate that the job should cease iterating. The default behavior of checking the interruption adapter and checking the runtime against a configured maximum could be elements in the chain of responsibility. Application authors could then add additional elements to the chain, like checking if a required resource is healthy, for example.