fix shim loop performance

[rpeltekov]

the shim loops in junction with the linear gradient did not seem to work out that well... Should probably write up a notebook making sure that output from each of the shim loops' is indeed linearly observed from the basis map and resulting fieldmap.

obtain a background first iterate through every loop on the current setup

• obtain a fieldmap with that single loop at currents (0.2- 1.6 amps, at let us say .2 am increments)
• subtract background from each of them
• show that they linearly correspond with one another. to do this use this sorta algorithm:

  for i in range (1,8):
    div = map[i] / map[i-1]
    div_avg = np.nanmean(div)
    div_std = np.nanstd(div)
    div_mode = np.nanmode(div)

• plot all these acquired statistics against the current that was applied. or maybe divide all of them by the lowest current and you should see a straight linear line in the mean and also low std...