Need to integrate async

[vsoch]

Right now we are issuing a command, waiting for the result, etc. We will need to be able to do this to launch a job /process and then come back to check it later. I want to develop the more core features of the library first.

[vsoch]

A few options:

• many examples with multiprocessinghttps://pymotw.com/2/multiprocessing/basics.html)
• I've had luck with creating a pool (e.g., gridtest)
• there is also os.fork which would be limited to linux, and the child process wouldn't exit with the parent.
• there is also asyncio

I think a lot of these solutions would work well given a constantly running server (e.g., the web interface) but for just the command line version, it would be more "run on demand."

[vsoch]

• https://curio.readthedocs.io/en/latest/tutorial.html#tasks-and-concurrency
• https://trio.readthedocs.io/en/stable/

[vsoch]

Picking up conversation from #24 - @yarikoptic time wise, I think this will be best to develop when we are actively working on an executor that warrants needing more time. I think we will want to:

• test a few strategies (different async libraries, for examples) and decide on the best one
• if it's appropriate to make as a default (using or not) then set this for the base executor, allow subclass executors to change the default if they deem fit
• if appropriate, allow the user to specify using one or the other (e.g., in gridtest I run things default in parallel, and give an option to run --serial which is useful for debugging.

I'd like to first merge #32, then address the logging and system exits, and then possibly we can get back to this when we develop an executor that warrants it (e.g., datalad or repronim).