instrument setup needs to be overhauled as it is UGLY

[benburningham]

perhaps ditch the FWHM flag, and instead supply an instrument table with:

• wavelength range
• convolution model (prism, R, dL, photometry*)
• convolution metric
• independent scale flag
• independent fudge flag
• others???

then can unpack in a loop (with fudge and scale in arrays for easy set in priors)

Makes it trivial to try different instruments and data combinations

• filters to be read in and stored in array as part of runargs , and number used as conv metric