Implement error-weighted reprojections

It would be great to be able to error-weight the interpolated and flux-conserving algorithms that would return a new science and error map. At this point, I just reproject the science and error images individually, which may or may not be the correct way to do this.

Start by inspecting the interpolated and flux-conserving reprojection entry points and how science and error images are currently handled separately. Clarify the intended error weighting and propagation, then define completion as both algorithms returning a science map and a correctly weighted error map with tests covering the results.

Written by the indexing model from the issue text.