Do not transport source particles below the energy cutoff

[GuySten]

Description

Source particles born below the energy cutoff for their type used to be transported until their first collision. Particles created during transport are checked against the cutoff when they are made, and existing particles are checked after each collision, but source particles were never checked. With this change, below-cutoff source particles are handled the same way as particles that fall below the cutoff during transport:

• Neutrons are killed in initialize_particle_track, with no energy deposited. This matches neutrons killed after a collision. Their weight still counts toward total_weight, so tally normalization is unchanged. They skip weight windows, so they can't be split. In event-based mode, dead source particles are no longer queued for transport (process_init_events).
• Photons get a collision distance of 0 in event_advance. They collide where they are born, and the existing cutoff check in sample_photon_reaction kills them and deposits their energy locally, which heating and pulse-height tallies score. In a void they are killed without depositing energy. This check applies at every step, so photons made below the cutoff by neutron reactions or atomic relaxation also deposit where they're made instead of travelling to a collision first.
• Electrons and positrons need no change: in a material they already collide and deposit energy at their birth site.

Checklist

• I have performed a self-review of my own code
• I have run clang-format on any C++ source files (if applicable). The local clang-format version wasn't checked against the required version 18.
• I have followed the style guidelines for Python source files (if applicable)
• I have made corresponding changes to the documentation (if applicable)
• I have added tests that prove my fix is effective or that my feature works (if applicable): test_source_below_energy_cutoff in tests/unit_tests/test_energy_cutoff.py

Note: this code has not been compiled or run locally, because the build environment had no HDF5. CI will be the first check.