Refactor metadata not flexible enough

[pmneves7]

In the NXcanSAS standard, certain metadata values are specified specifically that are officially part of the NXcanSAS standard. However, additional metadata may be optionally specified when looking at the inherited NX class. For example, the SASsample class has ~10 quantities within it in the NXcanSAS standard, but the inherited NXsample class can have about 60 additional quantities. Further, I can guarantee (because I have looked at the metadata from different instruments) that additional nonstandard optional quantities may be included that are useful to keep with the metadata.

For example, in the GP-SANS metadata, the omega and tilt_angle values are included in the sample metadata. These angles are physical motors that are highly useful to retain in the metadata, but are not part of any standard.

At least as it is implemented in temp_hdf5_reader in the refactor branch, the current dataloader does not look for any parameters not mentioned in the NXcanSAS standard. Further, the current metadata objects only have attributes for a subset of these parameters (and these parameters and where they are stored have moved in different cansas versions).

I suggest that all additional reported quantities in a given group (like SASsample) be included as procedurally generated quantities, copying the name, value, and unit from the hdf5 file reader. That way, additional metadata can be retained and exposed for experiments with unique equipment, samples, and sample environments.

This will also have to be re-evaluated once the cansas 2.0 standard is officially published.

Relevant links:
https://manual.nexusformat.org/classes/applications/NXcanSAS.html
https://manual.nexusformat.org/classes/base_classes/NXsample.html
https://www.cansas.org

[pmneves7]

We should also add the group name as metadata into each canSAS_class (sample, instrument, dataset, etc) because the group can have any name which may contain useful information. See for example:
https://www.cansas.org/formats/canSAS1d/1.1/doc/overview.html#xml-layout-for-multiple-experiments

[pmneves7]

Note that this is actually implemented as metadata.raw as a MetaNode type. It may however make more sense to put the relevant parameters into other parts of the metadata object. For example, arbitrary sample parameters could be added to the metadata.sample.