Consider astroalign for tracking corrections

[wtgee]

We currently plate-solve each image and do a comparison with the pointing image in order to determine the x and y pixel offsets that are used for tracking correction. This works but can be slow depending on the plate-solve, especially when operating from a Raspberry Pi.

Look into using the new astroalign library to obtain pixel offsets.

Describe the solution you'd like Replace the plate-solving in pocs.observatory.Observatory.analyze_recent with an astroalign offset.

https://github.com/panoptes/POCS/blob/develop/pocs/observatory.py#L492-L498

Related Issues/PRs #489