General loading of sample orientation

[jl-wynen]

This issue also pertains to the file writer and maybe simulation framework. But I didn't know where else to put it.

The current workflow loads the sample orientation using the a3, a4 parameters of a McStas simulation:

essspectroscopy/src/ess/spectroscopy/indirect/workflow.py

Lines 620 to 622 in fa79d4f

	# this is very BIFROST simulation specific -- can it be less so?
	# TODO use the _sample_ orientation itself to define a3 (and goniometer angles?)
	logs = _load_named(data['entry/parameters'], snx.NXlog, ('a3', 'a4'))

This won't work for 'real' data. As the TODO comment suggests, we should probably use NXsample. See in particular sample_orientation and orientation_matrix. I don't know how those are meant to be used or why there are two different fields.

We already have some related code: https://github.com/scipp/scippneutron/blob/3d6664b861da0b770fed4c06d9162323f48f4d4f/src/scippneutron/conversion/tof.py#L552

[SimonHeybrock]

Note that NXsample should be loaded correctly by ScippNexus if it contains such matrices.

[jl-wynen]

I talked to Greg about this offline. It seems unlikely that we will use the sample instead of a3, a4. The latter encode the orientation of the sample table not the sample itself. Data reduction only really needs the orientation of the table. And that will be encoded as some kind of log from NICOS. So that might well be in /entry/parameters which has class NXparameters.

[jl-wynen]

I am keeping the issue open until we have a clear decision.