Need underload value for detectors

[phyy-nx]

NXdetector specifies the saturation value for the detector:

saturation_value: (optional) NX_INT

The value at which the detector goes into saturation. Especially common to CCD detectors, the data is known to be invalid above this value.

The corresponding value, underload, is missing. Granted, in pixel_mask, bit 2 is classified as under responding. However, having a separate underload value available means a global number can be specified.

Note that in ImageCIF, underload is not explicitly defined, but it does define the parameters overload and undefined_value:
https://www.iucr.org/__data/iucr/cifdic_html/2/cif_img.dic/Carray_intensities.html
https://www.iucr.org/__data/iucr/cifdic_html/2/cif_img.dic/Iarray_intensities.overload.html
https://www.iucr.org/__data/iucr/cifdic_html/2/cif_img.dic/Iarray_intensities.undefined_value.html

So we could use undefined_value, but to me that does not imply pixels less than this value should not be included. If I had my druthers we would use trusted_min and trusted_max since those are very clear. However, since we already have saturation_value, I'd propose underload_value to match it.

Proposal:

underload_value: (optional) NX_INT

The lowest value at which pixels for this detector would be reasonably measured. The data is known to be invalid below this value.

The choice of language here matches the implied inequalities for saturation_value and matches the implementation in DIALS for generating a mask from a trusted range (min and max):
https://github.com/dials/dials/blob/master/util/masking.py#L168

mask = (imd > low) & (imd < high)

However, I think we could be more explicit in both saturation_value and underload_value by adding this sentence to both entries:

For example, given a saturation_value and an underload_value, the valid pixels are those less than or equal to the saturation_value and greater than or equal to underload_value.

[phyy-nx]

Approved during telco