Handle instrument calibration uncertainties for astrometry

[semaphoreP]

We should be able to marginalize over plate scale/north angle uncertainties.

[semaphoreP]

We need some way to recognize different instruments.

[sblunt]

Note that in PR #197 we introduced ability to recognize different instruments. This issue should be a lot easier to handle now.