have an image-based reference implementation

[stevenstetzler]

i would like to implement an image-based reference implementation that loads images from FITS files with an associated WCS. the "detection catalog" is a list of ra/dec associated with the center of each pixel in the FITS images. in this case, the coef value in search should likely be the flux value from the FITS file.

1. create a loader that converts a list of FITS files into a detection catalog using the ra/dec of the center of each pixel using the WCS stored in the FITS file with time information taken from the standard FITS header. flux values from the images should be stored in a column called "flux" in the resulting astropy table
2. modify the search methods so that coef refers to a column of the detection catalog instead of a fixed value. search methods utilize an N x 3 array of [ra, dec, time]. modify this so that it can take optionally take an N x 4 array of [ra, dec, time, flux]. perhaps add a coef_index value to the search and vote methods and modify the cpu / gpu implementation methods to use this coef_index of the X array instead of a single value.
3. create a search_image method that loads the FITS files, sets dx = 1 pixel scale (from FITS headers), generates directions from SearchDirections, and then calls search with the correct coef_index = 3