[LPD] [Dark] Save LPD bad pixel dark data as uint32, not float64

Description

The BadPixelsDark constants for LPD were being saved as float64, when they should be uint32. This isn't causing a practical problem, but it's unnecessary work for saving and loading constants, and wastes disk space.

Fixes calibration/planning#138

How Has This Been Tested?

TBD

Types of changes

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• My code follows the code style of this project.

Reviewers

@hammerd @schmidtp