using `DaskExecutor` is noisy

[cicdw]

dask.distributed is very verbose with its logging, which is a good thing; however, for newer users (especially those unfamiliar with dask distributed), this can be scary. We could potentially silence all logs that are not "CRITICAL" level by default, with a kwarg that allows this to be toggled during development.

Another set of logs come from dangling connections / streams and tend to look like:

tornado.iostream.StreamClosedError: Stream is closed
...
distributed.comm.core.CommClosedError: in <closed TCP>: ConnectionResetError: [Errno 54] Connection reset by peer
...

these are innocuous but annoying; it would be a good research task to dig into this and see if we can find the source, for our own peace of mind that we clean up our processes appropriately.

cc: @joshmeek

[jlowin]

@cicdw maybe this can be closed?

[cicdw]

Very very close; I'd like to know for sure what causes the following remaining warnings:

distributed.comm.tcp - WARNING - Closing dangling stream in <TCP local=tcp://127.0.0.1:65500 remote=tcp://127.0.0.1:65469>
distributed.comm.inproc - WARNING - Closing dangling queue in <InProc local=inproc://10.0.0.41/52568/123 remote=inproc://10.0.0.41/52568/116>

and see if I can make them go away, then I'll be satisfied. (You can reproduce these warnings locally with pytest -s).