Dont hang outstanding jobs when blockstoreManager shuts down

[dirkmc]

Fixes #237 by adding an error parameter to the job function. When the job is scheduled it first checks to see if there was a scheduling error before running.

[Stebalien]

I'm concerned that this just drops the jobs then does nothing. How about #240?

[dirkmc]

I think #240 will still hang outstanding jobs on shutdown, because in the worker loop it returns on px.Closing() (and doesn't drain the queue)

I'm concerned that this just drops the jobs then does nothing

At present we just ignore errors (and don't add anything into the results map for CIDs that get an error), which is why I was thinking that it should be ok to do the same if the process is closing.

[Stebalien]

I think #240 will still hang outstanding jobs on shutdown, because in the worker loop it returns on px.Closing() (and doesn't drain the queue)

It shouldn't, unless I'm missing something. There's no buffer so:

• All in progress jobs should execute.
• All unscheduled jobs should fail inside addJob.

[dirkmc]

Ah yeah I see that's true, I like this solution better.
Do you want me to remove the contexts as well?

[Stebalien]

We can keep them for now. That's more of a future "this is tricky" concern.