[fix/enhancement] move SN_models out of final_values

[maxbriel]

We've likely hit the maximum number of columns in a single HDF5 dataset for the final_values with the increased number of SN models.
As such, we need to separate out the SN_models to make sure we will not hit the same issue in the future.

I've rewritten the add_post_processed_quantities to:

1. Use the faster recfuntions from numpy to add all columns in one go to the structure array, final_values. Originally, we were writing one column at the time. deleting the existing final_values and then writing the new final_values with 1 extra column. Repeating this process for each new column we wanted to add (several hundred times given the number of SN_MODELS we have). This is a massive I/O overhead.
2. Store the SN_MODELS in their own group (grid/SN_MODELS) in the HDF5 file. Each SN_MODEL has its own dataset + in the attrs the parameters for the model are stored.
3. grid.compression_args are not actually populated correctly when loading a grid with grid(path_to_file) as such when writing the "new" datasets to file, they were uncompressed.

Caution

Bullet point 3 means that the final_values dataset in the release dataset are uncompressed!. I verified that this is the case!

Screenshots of uncompressed final_values

final_values

initial_values

This PR now also changes the IF_interpolator and interpolator. If required, the IF_interpolator now uses _get_grid_column to get the final_values or SN_models. It only does this if it looped over the SN_MODELS before.

Changes to step_MESA is just how we access the SN_MODEL data. It should be checked if this stored everything exactly the same as before. step_SN is able to access this correctly at the moment.

Warning

There is a bug I noticed with the LazyHDF5 accessing the HDF5 file when adding a column. When adding something to the HDF5f file, we reload the file, e.g. grid._reload_hdf5_file(). This invalidates the LazyHDF5 linking. We should fix this, if possible, but not in this PR.

[maxbriel]

Here are the output files from the binaries that I ran with the fix and v2.3. I originally posted these on Slack and forgot to add them here:

evolve_1Zsun_fix.log
evolve_1Zsun_main_.log

[sgossage]

We should keep this in mind as a to do item. (Not necessarily for this PR though.)