Refnx

------
- upgrade to PyQt6. PyQt6 is currently only available as a wheel from PyPI,
to be installed by e.g. pip. There are currently no conda packages for PyQt6.
PyQt6 enables the gui to be used on macosx_arm64, there are no PyQt5 wheels
available on that platform.
- further modification to use qtpy. qtpy is a shim that allows either
PyQt6 or Pyside6 to be used to run the refnx GUI. qtpy is a new dependency
for running the GUI.
- add Objective._generate_generative_mcmc to yield fit curves corresponding
to MCMC samples.
- add Structure._generate_sld_profile_mcmc to yield sld profiles corresponding
to MCMC samples.

-------
- event mode reduction for SPZ
- Display SLD profile for first structure in MixedReflectModel
- Github: changed the default branch from master to main. This is done to
match other repositories behaviour and to reduce cognitive load.
- Added a Parameter.corner method to plot the posterior distribution for a
given Parameter.
- Fixed a bug in MixedSlab

------
- Amend setup.py package discovery
- Change the manual beam finder box to green
- When a Parameters object contains a _UnaryOp or _BinaryOp the dependencies
of those objects are queried for varying parameters when asking for
Parameters.varying_parameters()

------
- First appearance of Parameter.set_constraint().
- Previously Parameter constraints were limited to algebraic Python
expressions. Now Parameter.set_constraint can be given a callable which
is evaluated at runtime, allowing more sophisticated constraints to be
developed. (N.B this is separate to the inequality constraints that can
be made during curvefitting using the NonLinearConstraint machinery).
- Add a Parratt recursion for calculating reflectivity, it seems to be a
little faster.
- Ported the pymc wrapper from pymc3 --> pymc (v4), which was released in
Jun2022.
- added sequence_to_parameters function.
- added LipidLeaflet.make_constraint.

------
- Don't pickle the manual_beam_finder.
- Fixed a bug when inserting a Component into a Stack in the refnx GUI. If
it was appended to the end of the Stack Component addition worked.
Inserting a Component into the middle of the Stack appeared to put it in the
correct place in the GUI, but inserted it in the wrong place in the
Structure.
- Added Lagrangian multipliers to GlobalObjective, allowing the user to
modulate the contribution of individual log-likelihoods to the whole. This
may be of use when co-refining Neutron/X-ray/Ellipsometry datasets.

------
- add name="refnx" to pyproject.toml.