Need to use bin centers in downsampling

[jradavenport]

Pointed out from @MelissaGraham:

There is a bug in the downsampling in pydis.DefFluxCal, the downsampling used bin edges instead of centers and skewed the sensitivity function.

Here is the proposed fix:

for i in range(len(std_wave)):
    rng = np.where((obj_wave>=std_wave[i]-std_wth[i]/2.0) &
                   (obj_wave<std_wave[i]+std_wth[i]/2.0) )
    IsH = np.where((balmer>=std_wave[i]-std_wth[i]/2.0) &
                   (balmer<std_wave[i]+std_wth[i]/2.0) )

here is her demonstration of this fix:

[jradavenport]

implemented in latest commit