garlic: implement support for systems with two coupled spins

[stestoll]

This is an occasionally requested feature. garlic currently does not support spin systems with more than one electron spin. If all lines have identical widths, then one can use pepper with isotropic g and A tensors. However, in the fast-motion regime, when linewidths depend on the transition (e.g. A + B mI + C mI^2), then garlic is needed.

Here are some relevant references (there are many others as well - see references cited in these papers):

1. Eaton et al, Metal-nitroxyl interactions VI. Analysis of EPR spectra of spin-labeled copper complexes, J. Magn. Reson. 1978, 32, 251 (link)
2. Sawant et al, Metal-nitroxyl interactions. 24. Electron spin delocalization in vanadyl and copper(II) bis(.beta.-diketonates), Inorg. Chem. 1982, 21, 1093 (link)
3. Eaton et al, Resolved Electron-Electron Spin-Spin Splittings in EPR Spectra, in: Berliner (ed), Spin Labeling. Biol. Magn. Reson. 1989, 8, 339 (link)
4. Eaton et al, Continuous wave electron paramagnetic resonance of nitroxide biradicals in fluid solution, Concepts Magn. Reson. A 2019, 47, e21426 (link)

Some thoughts:

• It needs to be determined how well the theory from these paper integrates with the second-order perturbation expressions implemented in garlic. If there are issues for second order (e.g. cross terms between J and A, J and g, etc.), one could potentially limit perturbation theory to first order in the case of J-coupled spin pairs.
• Fast-motion linewidths will likely contain some coupling-dependent terms, in particular dipolar coupling. Is the theory for this worked out somewhere?

[akboudalis]

Maybe the treatment here can be useful: https://www.tandfonline.com/doi/abs/10.1080/0026897021000054808