Refs #11641 Fix IntegrateMDHistoWorkspace

[eXeC64]

This is for trac ticket #11641

There is a rounding error in IntegrateMDHistoWorkspace that can wreck its output, i.e. cause it to discard data by looking at the incorrect bins.

To see this, run the following:

CreateMDWorkspace(Dimensions=4, Extents='-10,10,-10,10,-10,10,-10,10', Names='a,b,c,d', Units='u,u,u,u', OutputWorkspace='ws')
FakeMDEventData(InputWorkspace='ws', PeakParams='10000,0,0,0,0,1')
CutMD(InputWorkspace='ws', P1Bin='1', P2Bin='1', P3Bin='1', P4Bin='1', OutputWorkspace='ws2', NoPix=True)
IntegrateMDHistoWorkspace(InputWorkspace='ws2', OutputWorkspace='slice', P1Bin=[-1,1], P2Bin=[0.3,0.9], P3Bin=[-0.05,0.05], P4Bin=[-0.05,0.05])
plotSlice('slice')

You'll get a uniform output of ~149.5. It should be ~631.2

This will be much easier to see once #655 is merged, as you get obviously incorrect output. But for now, this seems to be the simplest way to reproduce it.

[martyngigg]

This seems to produce more sensible signal values now.