Lx scatter and Nikolic+2024 scaling relations

[daviesje]

Currently WIP PR for adding the xray luminosity scatter, I am currently testing the outputs to make sure we get what we expect.

Changes:

• Added another RNG field to HaloField and PerturbHaloField for the scatter in xray luminosity
• Added another field to the HaloBox and XraySourceBox to represent the xrays
• Added scaling relations for Metallicity, xray luminosity and mass-dependent SFR scatter to HaloBox.c
• Replaced the SFR field with the xray field in SpinTemperature.c for the xray calculations (lyman alpha / lyman werner still uses SFR)
• Refactored HaloBox.c to be more easily extendable and readable

[daviesje]

I've written a quick test which just compares the outputs of the scaling relations to the power-laws we expect over a few halo masses. In the meantime I was working on #404 and it made sense to combine the changes now (and it also avoided the need to expose the lyman werner turnover function or rewrite it in the test). A summary of these changes is below:

• Add a floor to the turnover mass for both MCG and ACG, controlled by the existing parameter M_TURN, which had the same function before but only worked when minihalos were switched off
• Add the option to simply extend the ACG scaling relations to MCG by setting the MCG parameters to None, this works by having conditional defaults. The defaults have been changed such that the SHMR is the same for both ACG and MCG. Due to our treatment of MCGs as 100% PopIII stars I have left the escape fraction defaults the same until we examine this model.
• Fixed a few bugs in the minihalo case and scaling relations

[daviesje]

The parameter default issues we discussed was fixed by 5759faf. Since we don't return the default structs anyway I simply do the validation after removing them instead of before. I'm not sure if it's worth a cherry-pick into master since it's just a quick fix.