Local data GFs should calculate to depth of current fault (or more)

[degoldberg]

-o add_data -st/--cgps assumes a max depth that is sometimes shallower than the deepest subfault depths. It throws an error when using get_strong_motion to re-compute GFs, but will still do inversion even without proper GFs. This should throw a fatal error and not allow the inversion to be run until GFs are recomputed