Plot options for surface plots from group workspaces

[tom-perkins]

Follows on from issue #14461, which added a right-click option to create a surface plot from a workspace group:

• [x] Context menu option not seen when workspaces are multi-spectra - set what spectrum numbers to plot from subworkspaces
• [x] Sometimes the menu option is seen for multi-spectra workspaces, but the resulting plot is empty. For example, POLREF00014966.nxs from the unit test data.
• [x] Rather than plotting against workspace index, allow user to specify name of axis (e.g. 'Temperature/C') and log ('temp-log') then extract value from log for each sub-workspace. Plot against this log value.
• [x] If the values are not recorded in a log, they can be added afterwards using the AddSampleLog algorithm. Provide a graphical way to do this.
• [x] Test performance using large datasets (100 runs, each 5000 channels)
• [x] Try to improve performance of plotting 3D graph
• [x] Support contour plots as well as surface plots
• [x] Deal with centre bins or histo workspaces (see comment below)
• [x] Enable more than one contour plot at a time
• [x] Update release notes

The linked Python files in the comment below show what the desired output should look like.

[tom-perkins]

The issue when the plot is produced, but is empty, occurs when the group contains only 2 workspaces (like POLREF00014966). Can replicate by creating a group with 2 sample workspaces and plotting.

The context menu option should be shown whenever the workspace clicked can be cast to a const WorkspaceGroup. I can't replicate the case when it is not shown.

[tom-perkins]

Performance: creating empty workspace and filling is much faster than using CreateWorkspace algorithm or plotting from a table. Tested on a group of 100 workspaces, each with 5000 points per spectra - took a few seconds to plot (~6), almost all of which was spent in the QtiPlot plotting code (plotGraph3D).

[tom-perkins]

These files were sent by Pascal Manuel as examples of the desired output (they work on the data in Dropbox). From his email:

As highlighted in my previous email, some workspaces are histo (this would be the case of a normal workspace) and some are centre bins (if you reload gss files). The algorithm should be clever enough to decide which type you want to plot and modify accordingly (this is the reason for the np.histogram in the files atached).

[tom-perkins]

@OwenArnold Adding so you can see the issue

[tom-perkins]

Left: QtiPlot, right: Pyplot Using colormap "Jet" and AddSampleLog to add log data, and ConvertToPointData on loading as described in email.