fit guess amplitude values when binning=True

[cjperks7]

When binning=False, it was pretty straightforward to choose guess values for background and amplitude bounds and guess values just looking at the y-values when plotting the data.

When binning=True, the user now has to interpret these values as an amplitude per bin, so when choosing guess values they take the y-value and divide by the bin width. But I'm giving spectrally a wavelength vector that's not uniform (it's technically the inverse of a uniform energy vector) which it's handling just fine. But since my bin width therefore changes it's not as intuitive to me what to take as the bin width, particularly for the background.

In general, this seems to be a can of worms that might repeatedly come up with new users so maybe harmonizing with the way picking bounds/initial guess for amplitudes when binning=False, keeping the rest under the hood.

[cjperks7]

Thining about it, I guess the current mentality is to account for the fact the data is technically counts/bin, but wouldn't the user's provided data already be in those units making the bounds/initial guess units already accounting for amplitude per bin using just the y-values?

But I also guess mathematically, the y-values are \int_{bin} (\flux [ph/s/AA]) *(d\lambda [\AA/bin]) = [counts/bin] for the histogrammed (discrete version of above integral) data I'm currently giving it