Improvements for parameterize/queues

[stefdoerr]

• Make names more sensible to distinguish them on queueing systems (i.e. for parameterize add prm at the beginning, or anything the user specifies?)
• Queue.wait should have an option for reporting current inprogress every x minutes. Dihedral jobs with thousands of angles/dihedrals look quite dead after a few days.

@raimis you think you could do number 1?

[raimis]

Ideally, the job name should identify molecule, dihedral, and rotomer index, but in practice queue systems limit job names to a few characters. I'll think about a naming scheme.

[j3mdamas]

The current name is automatically generated through _autoJobName. Example for LSF:

htmd/htmd/queues/lsfqueue.py

Lines 214 to 215 in 584eb2a

	def _autoJobName(self, path):
	return os.path.basename(os.path.abspath(path)) + '_' + ''.join([random.choice(string.digits) for _ in range(5)])

If we draw the parallelism with adaptive, where a single jobname is used for all epochs and replicas, we could do the same in parameterize cli:

SimQueue.jobname = args.filename.split('.')[0]

If one wants, we can also add -min, -esp, and -dihe to the end, if necessary to distinguish between all.

[j3mdamas]

For 2, here:

htmd/htmd/queues/simqueue.py

Line 75 in 584eb2a

def wait(self, sentinel=False):

we can add a report=None argument, which throws a logging message every report seconds, which we can set in the parameterize cli.

[stefdoerr]

For 2: #687

[stefdoerr]

#1 is very queue specific. Considering we have PM now we will just solve this there