Allow scattering code to trigger omega scans

[alurban]

This PR adds an option --omega-scans, whose argument is an integer, to gwdetchar-scattering. When --omega-scans N is nonzero, a random sample of N triggers that fall within active scattering segments is chosen to omega scan, with jobs launched in batch mode and displayed on the scattering page.

This PR also refactors the heavy lifting in gwdetchar-omega-batch into a new submodule, gwdetchar.omega.batch, including unit tests.

Test output is available here (requires LIGO.ORG credentials).

This fixes #287.

cc @duncanmmacleod, @siddharth101

[coveralls]

Coverage increased (+0.2%) to 95.512% when pulling f1ca592 on alurban:scat-omega into 9baa9d5 on gwdetchar:master.

[siddharth101]

Thanks a lot @alurban . This looks pretty good.

[duncanmmacleod]

One actual change, but more generally, can (and should) this block be moved into a function, to make it more modular, and unit-testtable.

[alurban]

@duncanmmacleod, that's probably a good idea, since there are now three tools (gwdetchar-omega-batch, gwdetchar-scattering, and hveto) that launch omega scans to condor. I'll have a go at that after I get back from lunch.

[alurban]

@duncanmmacleod, at the risk of scope creep, I've refactored the heavy-lifting parts of gwdetchar-omega-batch into a new module called gwdetchar.omega.batch and used this to launch omega scans from gwdetchar-scattering. This change includes unit tests for the new submodule.

Once it's released, we will also be able to use this to clean up similar functionality in hveto. I particularly like this because it removes a call to subprocess, it gives me the warm-and-fuzzies when we can do everything in memory.

[duncanmmacleod]

Looks like a great change, just one small thing to address.

[duncanmmacleod]

Can this be removed?

[alurban]

thanks @duncanmmacleod, I've removed it, this was left over from when I was writing the unit test and hadn't woken up enough yet

[duncanmmacleod]

LGTM.