Scanning 3DXRD Strain refinement

[jadball]

Summary:

We can now do point-by-point strain refinement for S3DXRD data, by implementing a method provided by @AxelHenningsson . Thanks for your help!
This works for grain maps coming from both the point-by-point indexing process and the tomographic mapping process.

PBPRefine:

This is a class to manage the point-by-point refinement process, similar to the PBP class.
Like the PBP class, we choose some peaks for self.icolf which we will refine with.
The class also has an input multi-valued map (self.pbpmap), a refinement mask (self.mask) and a single-valued 'choice' of the multi-valued map (required for computing x origins for each grain) (self.singlemap). For now this is computed from PBPMap.choose_best.
Features:

• Save and load from HDF5
• self.get_origins(): Compute origin-of-diffraction for each peak in self.icolf
• self.run_refine(): Initiate point-by-point refinement, either locally or on a cluster machine through SLURM
  Solves GrainMap class #278

PBPMap:

This is now a subclass of ImageD11.columnfile.columnfile. This solves the confusions with multi-valued data structures, ragged arrays, etc etc. in a very simple way that we were already using for the PBP refinement results

• Added plotting methods for nuniq histogram and best results (from notebooks)
• Added strain support (for strain refinement methods)
• Solves Multi-tensor map class #325

PBP:

• Now computes corrected origin of diffraction for g-vectors to account for x-position in the sample

Other bits and bobs:

• Updated Zenodo dataset with multiphase pars (updated fetch_data.py and its test accordingly)
• Renamed scanning notebooks to more clearly define the choice of processing routes (and how they diverge then merge together later)
• Added some more documentation to the scanning notebooks in the form of markdown text blocks
• Added some new utility functions to ImageD11/sinograms/geometry.py
• Added some more stress and strain functions to ImageD11/sinograms/tensor_map.py

TODO:

• Automated tests that run explicitly against @AxelHenningsson 's implementation (during development this was being done semi-manually)
• Test the functions I had to grab from ImageD11 to Numba-fy against their references
• Test stress stuff in TensorMap

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[jadball]

Python 2 tests are currently failing because roi_iradon is trying to import scipy.fft and failing

[jadball]

@jonwright are you happy to merge as-is and we exclude point_by_point tests on Python 2.7? I think this is the first time our CI touches the roi_iradon code which is why we haven't seen this before.

[jonwright]

Sure - go ahead and merge. I have the feeling that numpy V2 is going to break python 27 anyway. Probably it means skipping tests as I would like to keep a fitallb compatible code at least until we sort out the error bars on strain. Maybe that is quite close with this merge?

[jadball]

Got it, merging now! Managed to skip the test.