A round trip of SaveNexusProcessed and LoadNexusProcessed discards the detector table

[Dan Nixon]

This causes the same problems as observed with #10598: having spectra marked as monitors when they are not monitors causes analysis that relies on detector information to be completely incorrect.

This script demonstrates the issue (requires systemtests/Data on path):

Load(Filename='irs26176_graphite002_red.nxs', OutputWorkspace='sample')
SofQW2(InputWorkspace='sample', OutputWorkspace='sqw_original', QAxisBinning='0.6,0.1,1.8', EMode='Indirect', EFixed='1.845')
ConvertUnits(InputWorkspace='sqw_original', Target='Wavelength', EMode='Indirect', EFixed=1.845, OutputWorkspace='original_conversion')
SaveNexusProcessed(InputWorkspace='sqw_original', Filename='sqw_out.nxs')
LoadNexusProcessed(Filename='sqw_out.nxs', OutputWorkspace='sqw_loaded')
ConvertUnits(InputWorkspace='sqw_loaded', Target='Wavelength', EMode='Indirect', EFixed=1.845, OutputWorkspace='loaded_conversion')
plotSpectrum('loaded_conversion', range(0, 3))
plotSpectrum('original_conversion', range(0, 3))

[Dan Nixon]

Remove check that only saves detectors when V axis in spectrum no

Refs #10663

Changeset: 3560dd082adf0fa42a2b356b8c26f8d67af7f695

[Dan Nixon]

To test:

• Verify no unit or system tests were broken

• Run the script in the description
• Verify the plots are of the same data

Note that the loaded spectra map is not technically correct as the spectrum numbers are lost (which will be addressed in #10678), however it is far more important that the detector numbers are loaded and assigned to the correct spectra.

[John Hill]

Merge remote-tracking branch 'origin/bugfix/10663_savenexusprocessed_does_not_save_detector_map'

Full changeset: ​c305a447562b4c2c44a89b4ec26d8ba3d902baf1

[Stuart Campbell]

This ticket has been transferred to github issue ​11505