There is no longer any reason to use procrunner in place of built-in subprocess

* Unbreak help test
* explicitly set program name
* minor change to docstring

* Fix for nxmx import overwritten by local variable

Introduced by cctbx#597

Another case were bugs have crept in through lack of tests (see cctbx#470)

* news

* Rename newsfragments/XXX.bugfix to newsfragments/641.bugfix

pkg_resources is now deprecated, as importlib has covered
most use-cases since python 3.8. This removes the code
causing warnings in downstream tests.

Also suppress numpy 'product' warnings when running pytest.

Upstream dependencies (e.g. numpy, scipy) have now dropped Python 3.8
support in latest releases.

Add Python 3.11 support.

See also https://dials.github.io/kb/proposals/dc3

---------

Co-authored-by: Nicholas Devenish <ndevenish@gmail.com>

When setting panels by PHIL, merge multiple definitions first by id. This fixes cctbx#299.

Previously, a size (3,) array would have erroneously converted
to a 0x3 length vec3.

This is now explicitly forbidden; you must pass at least a (0, 3)
multidimensional array.

Fixes cctbx#439.

Partial support for Rigaku Oxford Diffraction format images.
---------

Co-authored-by: Takanori Nakane <nakane.t@gmail.com>
Co-authored-by: Nicholas Devenish <ndevenish@gmail.com>

This was previously done implicitly, but only when dials_regression
was present. This makes the test explicit, and works with
dials-data (so that it will work on e.g. CI with restricted access
to test cases).

Without this, slices sometimes needed to be indexed by their original
image index, even inside the slice.

The previous workaround in 67aab2 (cctbx#485) for negative offset was
reacting to the erroneous behaviour introduced by including batch
offsets in imagesequence slicing.

Without this, dxtbx did not work as a subdirectory in
a larger CMakeLists project.

The E-32-0017 Eiger moved to P13 on 2021-5-22, where the goniometer axis no
longer needs overriding.

Check for that date in the MiniCBF, and check for the serial matching the date
range.

Co-authored-by: Christian Becke <christian.becke@fu-berlin.de>
Co-authored-by: Nicholas Devenish <ndevenish@gmail.com>

Add new Beam class "PolychromaticBeam" for polychromatic/multi-wavelength/wide bandpass experiments.

---------

Co-authored-by: DiamondLightSource-build-server <DiamondLightSource-build-server@users.noreply.github.com>
Co-authored-by: Nicholas Devenish <ndevenish@gmail.com>

Co-authored-by: Takanori Nakane <nakane.t@gmail.com>

At the moment this just adds the additional datum and allows
serialisation etc. Old experiment files where this is not set are
interpreted as having Probe.xray. Some electron diffraction format
classes are updated here to record Probe.electron.

Co-authored-by: davidmcdonagh <davidnmcdonagh@gmail.com>

Fix bit depth for Eiger / NXmx for i19-2

Add bit depth check from image metadata as implemented on other Diamond NXmx
installations. Fixes dials/dials#2473

Data in NeXus can be 3 or 4 dimensional.
3D: Nimages by slow by fast
4D: Nimages by Nmodules by slow fast

Slice image_size and reshape the raw_data in these cases.

Co-authored-by: Richard Gildea <rjgildea@users.noreply.github.com>

    towncrier --name=dxtbx --version='3.16.0'

This behaviour has changed in cctbx, however has not been
backported to the release.