High-resolution transmission line profile w/ inhomogeneity

[HajimeKawahara]

I think we can model a high-R transmission line profile w/ a VMR inhomogeneity on a chord using exojax. These are just initial thoughts, should be divided into several separated issues:

1. auto-differentiable transmission radiative transfer
2. opacity computation of metal/atom lines, already raised in #17
3. mapping over planet radial velocity distribution along a chord w/ a GP connection. A mapping technique might help this in the Bayesian context, hopefully.
4. multiple phase fit, given a planet orbit

The number of lines is limited for this feature, even for a single line is fine. So, we can use the LPF, and even with multiple phase and velocity distribution on a chord, computation complexity is not extremely large for HMC-NUTS (honestly I just hope)

@astrostevanus

[astrostevanus]

That's very interesting, I will try it after I remove the Doppler shadow :)