feat: full LBL analysis in EventRates CI validations

[Charlotte-Knight]

Update the CI validations to introduce full fits, starting with the LBL analysis.

I added a step to the CI validations to mount cvmfs, and then used the files we stored there to run EventRates. This gets compared to a previous run of EventRates in the same way that the current limited EventRates CI works. I run with 10% downsampling so the workflow completes in a reasonable time.

I also took the opportunity to refactor and standardise how we build the Docker images.

We should prioritise getting the Atmospherics integrated as well, though I'd prefer making an issue for this and solving it in another PR since I'll have less time next week to continue with this.

I also think that the validations could be done better in a few different ways (also future work):

1. The downsampling is configured with a sed command which is awful but I think the best way to do it currently. The alternative is keeping a copy of the Beam sample configs that only have the downsampling as a difference. Or we could introduce downsampling as a cli option for the EventRates app, which is my preference, but requires more work.
2. The CI validations could be integrated into an app like they are in MaCh3 core. If this was done then we could run the validations locally before pushing.

[github-actions]

This is your first PR, thank you for contributing to MaCh3 DUNE!

[henry-wallace-phys]

This all seems sensible to me