First change allows plots to be created if only signal is included (no data or mc).
Second change adds a special case to give the poisson error for n=0. This is obviously not a good approach, and we should instead do the proper Poissonian asymmetric errors, but better than no error at all for n=0.
Closing because looks like 1st issue was already addressed without me realising and 2nd probably not wanted in most circumstances - may add 2nd to another update.
First change allows plots to be created if only signal is included (no data or mc).
Second change adds a special case to give the poisson error for n=0. This is obviously not a good approach, and we should instead do the proper Poissonian asymmetric errors, but better than no error at all for n=0.