[feature] distributions enhancements

[drobnyjt]

trafficstars

This issue is going to take over from issue #62 once feedback on the format has been received. It will cover the various improvements I am considering for the distributions experimental feature.

• instead of binning, what about linear or more advanced weighting? It would produce much smoother distributions with no loss of fidelity; in fact, it would encode more information in them (although they would stop being integers and start being floating point arrays).
• is toml the appropriate output file format?
• add options to modify the reflected and sputtered energies separately; this is because the peak sputtered energy is ~Es, whereas the reflected energy spectrum often peaks at 90% E0