inserting diffusion-code in EPG

[mfuderer]

Next to pulses, relaxation, inversion and regrowth, we can model the effect of diffusion

[oscarvanderheide]

Nice. Before merging, I have some questions/requests.

1. Would it make sense to rename D to ADC?

2. Please add tests (in test/runtests.jl) that at least check:

   • When the diffusion coefficient is zero, that the simulation results are the same as if there was no diffusion coefficient in the parameters

   • When the simulations with diffusion effects run on both cpu and gpu, that give the same (in an \approx sense) results.

3. Also in src/interfaces/tissueproperties.jl, new variants of lines like const T1T2 = T₁T₂ can be added.

[oscarvanderheide]

Note to self: the EPG diffusion code seems to run fine on CPU. For simulation CUDA devices, the ConfigurationStatesSubset struct needs to be modified to include the configuration state indices it is holding.

[oscarvanderheide]

Another note to self: precompute diffusion decay matrix rather than recomputing its entries each TR