Some changes for more flexibility downstream

[benbovy]

Those changes are needed to fully leverage fastscapelib-fortran in fastscape while enabling more customization.

I exploit the fact that all variables declared in FastScapeContext are exposed to python (i.e., from fastscapelib_fortran.fastscapecontext attribute), which allows me to do more customization than what is possible by using the API (FastScape_api.f90).

[benbovy]

@jeanbraun when you have time could you review these changes?

It addresses #24. There is no breaking change. The only change in the API is when a negative value is set for g1 (aka g in the documentation), the array values won't be set unless accessing directly to the FastScapeContext. Users should never do that, but I hack this way from Python (xarray-simlab wrappers) to have more control on the array values. From the API point of view, I guess it is fine since the bedrock dimensionless deposition/transport coefficient should always have a value >=0?

I haven't done any deep checking, but everything compiles and the examples look all good.