Speed-up diffraction focusing

[Laurent Chapon]

Diffraction focusing is slow, and becomes a real problem for large instrument/slow machines. Should be able to take advantage of the focusing method to create only one vector temporary instead of chaining existing algorithms.

[Laurent Chapon]

(In [2168]) re#498 Add clear calls for the ynew and Enew vectors of the rebin routine to make sure reference to prevous data is erased.

[Laurent Chapon]

(In [2169]) re#498 First commit of this algorithm. It process each spectra at the time, which avoid large memory consumption when dealing with large datasets. Also it validates the consistency of the grouping in the file and the eventual Grouping of detectors in the workspace. The d-spacing range is now individual to each group and not common to the entire workspace. To do: validator for the Input workspace To do: propagate the geometry gouping. To do: find a way to weight the contribution of each spectra to each point in the range.

[Laurent Chapon]

(In [2177]) re#498 Minor: Improve message when throw non valid bondaries

[Laurent Chapon]

(In [2187]) re#498 Create a new function for rebinning. This is about 35% faster than old routine. Need to write unit for that. Speed-up diffraction focussing by a factor 2.5. It does take advantage of this new rebinning routine and also avoid copying each vector and only propagate errors at the end. On a 2GHz processor, running diffraction focussing on HRPD data now takes ~0.2 sec

[Laurent Chapon]

(In [2189]) re#498 Add workspace Validator for RAW counts and X-unit as dSpacing

[Laurent Chapon]

(In [2199]) re#498 Change to the new RebinHistogram. Should now be correct.

[Laurent Chapon]

(In [2216]) re#498 Change rebinHistogram. Was a mistake in upper bound

[Laurent Chapon]

(In [2217]) re#498 First part of the unit-test of new Rbin function

[Stuart Campbell]

This ticket has been transferred to github issue ​1346