'sim' cal_type

[jpober]

We're planning to use UVCal objects to store mock gain errors. Technically, these aren't calibration solutions, since they weren't derived from any calibration algorithm. (The goal would be to test calibration algorithms on their ability to recover these gains.) Seems like a new cal_type variable is warranted to store these data products, but I'm open to other solutions.

[jpober]

Should actually be supported through 'cal_style' not 'cal_type'.