Threshold MD always copies data, use in place changes if possible

[Nick Draper]

Using the fly dataset the memory use peaks at 30GB (double the datasets 16GB) even when overwriting the original dataset. Also this leads to a long delay before the algorithm progress starts.

[Owen Arnold]

Tester: The following should show these inplace operations working.

Show the original data.

CreateMDHistoWorkspace(SignalInput='1,2,3,4,5,6,7,8,9',ErrorInput='1,1,1,1,1,1,1,1,1',Dimensionality='2',Extents='-1,1,-1,1',NumberOfBins='3,3',Names='A,B',Units='U,T',OutputWorkspace='demo')
plotSlice('demo')

Threshold in place.

CreateMDHistoWorkspace(SignalInput='1,2,3,4,5,6,7,8,9',ErrorInput='1,1,1,1,1,1,1,1,1',Dimensionality='2',Extents='-1,1,-1,1',NumberOfBins='3,3',Names='A,B',Units='U,T',OutputWorkspace='demo')
plotSlice('demo')
ThresholdMD(InputWorkspace='demo',Condition='Greater Than',ReferenceValue='1',OverwriteWithZero='0',OutputWorkspace='demo')
plotSlice('demo')

Also try this out on the fly.vtk dataset if you can, as this was mentioned in the ticket description.

[Owen Arnold]

refs #7282. Modify in place and fix warnings.

Changeset: 3767e380bb1293d378212a477671f6110c8ee73f

[Owen Arnold]

refs #7282. Comply with OpenMP restriction on windows.

Changeset: f576c82aabcc2358877058bb5a08f5190e3c9d11

[Nick Draper]

works well, does not use any additional memory when thresholding in place and is much fater for it.

[Stuart Campbell]

This ticket has been transferred to github issue ​8128