Previous refactoring changed the last x value, did not seem to make a
difference though.

There was a bug in determining to total x-range on an empty event
workspace, and another bug in resampling from negative to positive
xmin/max with logorithmic steps (not mathematically possible).

Turns out that ordering the set is all that is needed to get
the numbering of output workspaces to match the splitters
information.

Refs #16689.

re #16600

I reviewed the documentation and there is no need for widespread changes to describe a multi selection list box.

This test passed locally and on PR builds, but broke the master
Win7 build. It may be unstable on certain machines.

Remove for now and come back to it at a later time.

re #16634

re #16600

re #16600

re #16600

re #16142

The algorithm loops over workspace indices. The "detector" column of the
table is filled with spectrum numbers, not workspace indices.

re #16142

Each thread works on a single spectrum at a time, so this seems to be
unnecessary.

- Change supplied frequency to MHz
- Add a test for frequency units
- Test detector IDs in table
- Update doc test to test that the detectors are numbered correctly in the table

re #16142