Null poissonian

[DBAnthony]

The minimum error of any bin in n should be 1.15, which is the poissonian error for 0 counts. We don't want this error to be plotted for every bin in our standard plots, but it can be turned on in the config with:
is_null_poissonian: True.

We won't want this for standard plots like below, but is useful for e.g. mountain range plots, or fit-binned plots. Obviously better solution would be explicitly calculating the poisson error and plotting asymmetric errors, but this is better than nothing. I'm taking the quad sum of 1.15 and the 'default' error so that the error is > 1.15 for nonzero yields. Ideally would find a way to ignore the underflow bin.

I'm also not sure whether this minimum value should be applied to monte-carlo, as I am currently, or just data.

With and without this option, and also with autoscaling also turned on, which looks better: