Fix a bug in time cutoff behavior

[GuySten]

Description

Currently, when particles are killed because of time cutoff they overshoot their final position and then moved back to the correct location but the scoring distance is not decreased.

This PR contain some of the changes proposed in #3525 that should be merged even if #3107 is chosen over #3525.
When this PR is merged I am planning to update #3525 accordingly.

Changes

• This PR change the logic of particle distance calculation in event advance so there is no overshoot.
• This PR also apply some fixes to multigroup mode to support calculation of particle's speed and time.
• This PR also change the formula for relativistic particle speed calculation to one without precision errors around small energy.
• This PR also fixes the time cutoff regression test which has unphysical results:
  Because of said overshoot, the particle tracklength in the sphere is the full 200 cm sphere radius instead of the ~138 cm that it should travel in the 1e-7 sec timeframe before the time cutoff.

Checklist

• I have performed a self-review of my own code
• I have run clang-format (version 15) on any C++ source files (if applicable)
• 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)

[paulromano]

Thanks a lot for separating out these changes @GuySten. For posterity, I thought I'd just write down the formula that you've implemented here for the velocity that overcomes the round-off error:
$v = c \sqrt{1 - \left( \frac{m}{m + E} \right )^2} = c \frac{\sqrt{E(E + 2m)}}{E + m}$

@sethrj @amandalund pinging you guys since the original formula in OpenMC was borrowed from celeritas (which looks like it is still using the same form in ParticleTrackView.hh). I would recommend adopting the above formula which is very robust against round-off error