Darknews

Features:
* Made `pyhepmc` an optional dependence, similar to `parquet`.
* Updated and added Example 4 and Example 5.

Bug fixes:
* Do not specify random number generator for decay_case instance (as opposed to MC_case). This crashed SIREN.

Bug fixes

* Fix compatibility issue between numpy2.0 and Cython modules. Specifying numpy array type explicitly and using `np.import_array()`.
* Fix CYTHONIZE flag in build -- it was incorrectly skipping cythonization because of an old CYTHONIZE os flag that was never set.
* Fix bug that crashed event generation for Booster beam experiments when neval < 100.

Tests
* Significantly reduced the cost of running tests. The prediction test now uses only 100 events.
* Test of ModelContainer and summary_plots()

Bug fixes

* included axial-vector coupling when calculating the Z' decay width
* Corrected a spurious minus sign in 2-body decays of Dirac HNLs.
* fixed a bug that was ignoring the neutron-elastic component.

Updates
* Allows epsilon_Z parameter for Z-Z' mass mixing. Not physical yet.

Bug fixes
Fix factor of 2 in dipole coupling
* When setting the transition magnetic moment parameters, e.g. mu_tr_mu4, DarkNews was artificially interpreting the coupling as ''d'' instead of ''mu_tr'' (as the name suggests). The relation between these parameters is d = mu/2, so couplings were artificially twice as large as they should have been (and rates four times as large). This bug fix makes the rates consistent with the calculations in [this paper](https://arxiv.org/abs/2206.07100).

Other fixes
* fix bugs in setting the logging level
* fix missing factor of 3 in three_charge in pdg.new_particle()
* update pyhepmc.io import according to the latest pyhepmc version.

Improvements
* delete ToyAnalysis. This will be replaced by the fastbnb package
* improved and updated examples
* allowed setting parameters to None in input files
* created CKM matrix
* created MiniBooNE RHC dirt mode preset experiment
* determining model type now within GenLauncher
* created input tests

**New features:**

1) New dirt implementation for BNB experiments. Uses a toy simulation of the decay pipe to sample in meson decay position (only for BNB detectors).
2) HepMC formats are now working for single photon final states.
3) The user can now set the desired sparseness of the final output. This change affects the internal dataframe in DarkNews before printing, so can also save a lot on memory usage.

**Bug fixes:**

1) Evaluating the form factor for Q2 ranges at large Enu (>>10 GeV) energies sometimes overflowed because of the smallness of the sinc functions. Cut-off in form factor for large Q2 values is now implemented.
2) Several bugs in the previous dirt implementation were fixed.

Version used by Mark to generate MicroBooNE benchmark points.


* Better dirt models for SBN experiments.
* Fixed model le_norm_state bug fix.
* MicroBooNE TPC-only experiment file.