resolution calculation does not check for basis of support between q points

[pkienzle]

To compute the resolution approximation more accurately all measured q points should have enough calculated q points within the resolution window regardless of how sparsely the data is measured.

sasmodels/resolution.py only extends q beyond the measured end points, it does not check that there is sufficient number of points within dq of each measured q point. This is generally not a problem for SANS since the 1D pattern is computed from the circular average with dense enough spacing, but it could potentially be a problem when datasets from non-overlapping q regions are joined, or when data is excluded from the middle of the q range. It could also be a problem in USANS where q spacing is completely under user control.

[pkienzle]

Yet another user wanted to fit a model with an excluded region in the middle. The GUI does not support masking central parts of the q range, and the resolution calculator does not fill the masked range with q calculation points, leading to poorly fitting points at the edges of the masked region.