Fitting with counts/bin width normalised data

[Nick Draper]

The interaction with fitting is confusing as you setup the plot against normalized data, but it fits against the underlying non normalized data.

This script creates test workspaces

uniform_bins = CreateSampleWorkspace()
non_uniform_bins = Rebin(uniform_bins, Params="0,300,10000,500,20000")
point_data = ConvertToPointData(uniform_bins)
non_uniform_bins_distrib = non_uniform_bins.clone()
ConvertToDistribution(non_uniform_bins_distrib)

[Nick Draper]

(In #10639) Reopened due to a few issues: only the first should be dealt with in this ticket, the others have new tickets.

1. If you plot errors, whether by plotspectrum with errors, or the graph->add error bars -> from workspace, they are not scaled by the bin width and therefore overpower the data.
2. The interaction with fitting is confusing as you setup the plot against normalized data, but it fits against the underlying non normalized data. - Follow on ticket to deal with this #10668.
3. We need to make sure the default is set "correctly" for this release - Follow on ticket to determine this and implement #10669.

[Roman Tolchenov]

Re #10668. Fit browser passes an option to Fit to normalise data.

Changeset: 9efbf2c9e4dd496246989169ab9073945b7d798d

[Roman Tolchenov]

Re #10668. Unit test for data normalisation.

Changeset: fee637561e08cb4430307671a61ee581a93b704d

[Roman Tolchenov]

To test

1. Run the attached script to create a non-uniform test workspace.
2. Fit the data using the graphical fitting tool with normalisation on and off.
3. Check that fitting works correctly in both cases.

[Roman Tolchenov]

Re #10668. Fix doxygen warning.

Changeset: 0ec3dc54ffdee84f002bf917b7c4f061b63c24bd

[Martyn Gigg]

Merge branch 'feature/10668_fitting_normalised_data'

Full changeset: ​27ffc73d3634b9ae3998dde61a0180ef24f6f7a9

[Stuart Campbell]

This ticket has been transferred to github issue ​11510