Update docs for unbinding timing changes + misc

[VictorForouhar]

This PR adds docs for the changes in #121, and some example plots of how unbinding time scales with number of particles (neat!).

I have also restructured the docs a bit. I thought it was weird to talk about the timing.log file in the diagnostics instead of the Output section of the code. Hence, I moved it there.

I also noticed a few additional things that should be addressed:

• Add docs for ReassignParticles parameter.
• Timers for read_snap and read_halo are currently meaningless. For example, read_snap is stored immediately after snap_hash.
• Some timers are swift specific (e.g. halo_comms and halo_io).

[VictorForouhar]

The timers are now placed on similar places across I/O formats, and I have removed those which would always measure 0 seconds. gadget binary format does not require communication of halo particles, unlike apostle and swift formats. Hence, there is not halo_comms timer for gadget.

I also removed the GSL parts of the code, which we have never used and were there to compute the eigenvalues of the inertia tensor. I think there are handier ways to do so, especially if we already provide the inertia tensors.

[VictorForouhar]

Updated documentation is in https://hbt-herons.strw.leidenuniv.nl/new_version/.

Main changes located in the following pages:

https://hbt-herons.strw.leidenuniv.nl/new_version/outputs/timing/#subhalo-level
https://hbt-herons.strw.leidenuniv.nl/new_version/outputs/subhalo_properties/#analysis-cost-statistics
https://hbt-herons.strw.leidenuniv.nl/new_version/installation/#debugging-and-testing