Explore mechanism to calculate difference in prim/sec signatures as the 'signature-metric'

[samland1116]

previously part of #523 but due to scope creep it will be it's own issue here

• In general, our signature_metrics are more analogous to just signatures
  • We need to devise a pattern to calculate the signatures (i.e. our existing signature_metrics), and compare those signatures between prim/sec (i.e. true signature_metrics)
  • Note that we still want the ability to return signatures as we do currently within this new pattern
• Some of this functionality -- like the ability to calculate signatures directly from the primary/secondary timeseries tables -- should be tackled in v0.6-dev. However, the pattern could be explored and implemented in v0.5 (main) using the joined_timeseries table.