feat(compactor HS): add support for worker for processing of jobs from the compactor's job queue

[sandeepsukhani]

What this PR does / why we need it:
It adds two new components related to making the compactor horizontally scalable:

1. Worker: It gets the job from the job queue over GRPC, executes the job, and sends the job response to the job queue. It keeps going while it can get jobs from the GRPC connection. If the stream gets disconnected, the worker tries to re-establish it to work on the jobs from the queue.
2. Worker Manager: It creates a specified number of workers and implements a Stop method to stop all the workers at once.

I have done some minor changes to the existing Queue code, like:

1. Retry the failed jobs in the goroutine we run for retrying the timed-out jobs.
2. Removed unused Stop function.
3. Close the q.stop channel before closing q.queue so that builders stop first, and they avoid writing to the queue, which otherwise would lead to panics.

Checklist

• Tests updated