Support multiple present background worker implementations

[plasticine]

For various reasons we support several background job implementations in production — namely Que and Faktory.

We’re very interested in job-iteration, however an immediate issue we face (other than support being added #88 & #89, which I’m happy to do) was the fact that the gem attempts to discover and configure the interruption adapter up-front for a given worker — here. Unfortunately for us given our multiple queue implementations in play that share a codebase/monorepo, it means that we’d hit this branch, and be stuck.

I think supporting multiple background job implementations, while perhaps a little unusual, is a very useful use-case to support. Supporting multiple implementations allows for a lot more flexibility around migrating from system to system, or in cases where multiple background systems are desirable in order to place jobs on background systems with differing characteristics.

I would propose that instead of configuring the interruption adapter once during init for a given worker we just discover it around the execution of each iteration, I think this would be fairly achievable and am happy to take a look if this is a patch that would be of interest. 😃