Use monitor normalisation functions from ESSreduce

[jl-wynen]

Requires scipp/essreduce#286
Fixes #167

[SimonHeybrock]

The linked issue also talkes about vanadium binning and data binning not matching. As far as I can tell this is not being prevent here, right? Especially for the (typical?) case of scientists wanting to store and reuse processed vanadium from a file.

[SimonHeybrock]

Is this now required to fix the bug? Does one have to use event mode in both or neither for correct results?

[jl-wynen]

It's not required. The fixup PR I made in ESSreduce addresses the histogram mode here. But using a histogram is less precise. In this case, we use lookup to get the monitor values with the bin centres of the detector. If the monitor has a finer binning than the detector, that will lead to ignoring many monitor bins.

[SimonHeybrock]

Wondering why we have the graph as a separate provider instead of lining? Seems simple and flexible enough?

[jl-wynen]

Because we do the same with the detector graph. I figured it's ultimately simplest to keep it symmetric.

[SimonHeybrock]

But it exposes implementation details (using transform_coords) in the compute graph.

[jl-wynen]

Is that a problem? We seem to be fine with it for the detectors. And it gives a customisation point if we need it.

[jl-wynen]

The linked issue also talkes about vanadium binning and data binning not matching. As far as I can tell this is not being prevent here, right? Especially for the (typical?) case of scientists wanting to store and reuse processed vanadium from a file.

Not the issue I linked. That only mentions monitor bins. Are you referring to #168 ? There are tests in ESSreduce that check that the result is independent of the monitor binning.

[SimonHeybrock]

The linked issue also talkes about vanadium binning and data binning not matching. As far as I can tell this is not being prevent here, right? Especially for the (typical?) case of scientists wanting to store and reuse processed vanadium from a file.

Not the issue I linked. That only mentions monitor bins. Are you referring to #168 ? There are tests in ESSreduce that check that the result is independent of the monitor binning.

I mean this:

if it is ensured that (1) the vanadium reduction uses the same monitor bin size

Is the new approach addressing this, since it is now not sensitive to the monitor bin size (assuming bin size is reasonable)?

[jl-wynen]

The linked issue also talkes about vanadium binning and data binning not matching. As far as I can tell this is not being prevent here, right? Especially for the (typical?) case of scientists wanting to store and reuse processed vanadium from a file.

Not the issue I linked. That only mentions monitor bins. Are you referring to #168 ? There are tests in ESSreduce that check that the result is independent of the monitor binning.

I mean this:

if it is ensured that (1) the vanadium reduction uses the same monitor bin size

Is the new approach addressing this, since it is now not sensitive to the monitor bin size (assuming bin size is reasonable)?

Correct. The implementation this uses is not sensitive to the monitor binning assuming a sensible binning.