HFile select function updates

[GarethCabournDavies]

The select function in HFile should be able to use any sngl-ranking statistic directly.

We should also be able to return indices-only, rather than just values, or indices-and-values as is the case now

To use the above selection, we should make sure we are efficient with the code, not reading in additional datasets which are not needed.

• I think we could do this by having a dictionary in ranking.py which stores which datasets are used by each statistic. That list can be used as a reference in this function.

mask_loudest_to_n_triggers should have the same functionality, acting on triggers rather than the file though

[GarethCabournDavies]

Ian pointed out that sometimes datasets aren't made if they aren't requested in the inspiral jobs - the 'used_datasets' dictionary should be able to handle this, but we should make sure of this

[GarethCabournDavies]

This is fully implemented now, so can be closed