ctools

Currently, it is possible to use ctselect to cut on a certain energy range (say 1-10 TeV). If the user makes this cut and forgets about it, it is possible that she/he applies another cut (say 0.5-50 TeV) later. The header keywords are updated accordingly and, more important, the GEbounds-definition. The ebounds definition is then used in the likelihood fit to apply the energy range.
The problem now is that the user has already thrown away every event outside [1,10] TeV. Accordingly, the likelihood fit does not find events outside this range. However, the IRFs might indicate high sensitivity in that region (the fitter thinks we are analysing from 0.5-50 TeV). Accordingly, the fit might get mixed up between zero source flux in the low energy range and signal between 1-10TeV.
The problem is the same for the RoI and GTI cuts.
Accordingly, ctselect should be able to read previously applied cuts and only cut further if the new cuts are contained inside the old ones.