ENH: Auto calibrate detector

[CJ-Wright]

I am not certain that this will work, but it may be possible to auto-calibrate the detector if the images meet certain criteria.
The basic idea is to make a poni with approximately correct beam center (and wavelength but we should get a precise value for that elsewhere) and zero tilt angles. Then use/abuse the pyFAI-recalib routine to correct our "calibration" with the tilt angles and such. This may eliminate the need for humans to click on rings, which would be very nice.

Caveats

1. We need the beam's position on the detector. This may be handled by some code from scikit-beam which was able to find the center if there were full rings on the detector (a common feature of PDF experiments at XPD and other high energy beamlines). Otherwise we could have the user click on the center, or pull it from a configuration file for stable detector/beam positions.
2. The "guess" of zero tilt may need to be fairly accurate, I don't know how much deviation pyFAI can handle before it starts to go off the rails. However, I don't that we can even integrate images which have significant deviations from normal incidence let alone calibrate them so this may not be a big issue.

[CJ-Wright]

See https://github.com/scikit-beam/scikit-beam/blob/master/skbeam/core/calibration.py#L142