change number of DM trials as function of dt

[caseyjlaw]

Currently we have a fixed DM grid that is searched for each resampling scale. This is clearly suboptimal, since larger values of dt require fewer DM trials.
Not clear how to use dm indices if we change the dm grid for larger dt. One possibility is to take a subset of dm that are approximately correct for larger dt, so:

dmarr = [0, 10, 20, 30, 40, 50, 60]
dt = 1
for dmind in dminds:
    ...search...
dt = 2
dminds = [0, 2, 4, 6]  # or whatever
for dmind in dminds:
    ...search...

[caseyjlaw]

Or at least skip trial if delay calc is equal to some previous calc?