REFERENCE: fixes to the some broken statistics

[GarethCabournDavies]

THIS SHOULD NOT BE MERGED

This is a reference PR to show the changes needed to fix statistics which are broken.

These are fixed in another PR, but I wanted to show which changes are specifically for fixing the existing statistics.

[GarethCabournDavies]

This is a reference issue, simply to point to the fact that the comparisons in #4608 are not against master. These are minor fixes and rare use cases, which can be subsumed into the changes of that PR

• The statistics which didn't use a single_dtype containing 'end_time' (mainly snr-like ones) were broken for singls_findtrigs , which assumed they would - we never used this so it hasnt shown up until I was doing full comparisons
• The logic was wrong on coinc_lim_for_thresh for the PhaseTDstatistic - again this statistic hasn't been used in offline searches so wasn't a problem before
• PhaseTDstatistic's rank_stat_single was not using the snglstat key for single info, which it should have
• ExpFitCombinedSNR's rank_stat_single was using the snglstat key for single info, which it shouldn't have
• ExpFitStatistic, PhaseTDExpFitStatistic and ExpFitBgRateStatistic look to be ones which are used as parent classes only, and don't work directly, so I've commented these out. These are completely removed in Update rates-based statistics to be modular #4608