Expand plutonium dioxide benchmark

[ElliottKasoar]

Just to follow up, I think for a section in the paper it would be nice to have a slightly more complete section on Actinide Dioxides (industry relevant actinide dioxide are ThO2, UO2, PuO2).

There are essentially 3 metrics that would be nice here.

1. Lattice expansion (experimental reference). Constraint: long benchmark, will do a production run and store results for each model.
2. Neutral defect formation energies (DFT+U reference). Code is equivalent for each cation. Constraint: Running own DFT+U for reference data (not done for UO2 and ThO2), can use reference papers for this if necessary.
3. PuO2 DFT+U parity: energy, force, stress, Done.

Again, should be quite easy (with AI tool support) to transfer existing code for these benchmarks to ml-peg format.

Originally posted by @williamdavie in #396 (comment)

We also need to double check the thresholds make sense.