Write flat and dark field data to the CIL Nexus format

[lauramurgatroyd]

Closes: #934
Writes dark and flat field data out to a CIL nexus file, and reads this in with the NEXUSReader.
The dark and flat fields must be arrays with the same dimensions and ordering as the data set.

Dark field data is saved to: entry1/tomo_entry/data/dark_field/data
A position key for this data is saved to: entry1/tomo_entry/data/dark_field/position_key
And similar for flat fields.
Any suggestions on renaming the location of these would be welcome.

The position key is an array of length [number of flat fields] which contains 0s and 1s. 0 indicates the field was taken before the projections, 1 indicates after.

I originally tried adding the flat and dark field data into the array that is saved in entry1/tomo_entry/data/data and saving an image key to identify which are flat fields, dark fields and projections (as is done in the NXTomo format), but this quickly became complicated due to different orderings the data dimensions could have (see here 86f2ae2)
It seems better to save the dark and flat fields as separate arrays because that is how they will be read back and used in processing anyway.

More detailed information about the ordering of the dark and flat fields (i.e. if some were taken in between projections) could be added to the nexus file in the future if needed - this could be in another array entry within entry1/tomo_entry/data/dark_field/

[lauramurgatroyd]

This may mean we don't need

CIL/Wrappers/Python/cil/io/NXTomoWriter.py

Line 29 in 786db5f

class NXTomoWriter(object):

as we would prefer to write out the CIL Nexus format containing the dark and flat field info, instead of an NXTomo file

[lauramurgatroyd]

To do:

• Add checks for dimensions of dark and flat fields relative to the data
• Add docstrings
• Add default values of dark and flat keys - or throw an error if they are not set

[lauramurgatroyd]

Closing this, because instead of writing dark and flat fields to the CIL Nexus file, we are going to keep the data as NXTomo format until we get to a processing step (e.g. Centre of Rotation Correction) where we need to specifically write out CIL info (such as the geometry). The other steps just modify the data which will be saved in an NXTomo anyway.
Therefore we will just need the NXTomo reader and writers (see #939 and #949)