add Beale's ratio estimator as its own model

[aappling-usgs]

not yet sure this is a great idea, because (1) BRE only operates at annual to multi-year timescales and thus ignores a lot of loadflex functionality, and (2) the BRE method is getting developed further this year, so it might be better to wait. still, worth considering b/c BRE has great performance at sites without trends when the target is multi-year means