Rework of Grackle

Rework of Grackle with version change (v2 to v3).

Now the code is a lot cleaner. I think if you wish to give a first feedback, it is a good moment.

The code currently works only with the mode 0 (interpolation from a table with a very basic chemistry). In comments, there is already a beginning of the future implementation for mode 1,2 and 3.

My next steps will be to implement mode > 0 and add some parameters in the yaml file. Do you wish to wait on theses for the merge?

Reproduction of Figure 2 in https://arxiv.org/abs/1610.09591.

Reproduction of Figure 8.

I am still getting some differences with the article, but I believe it comes from a different mode.

This graph is done with the CoolingBox test (the analytical solution is for a constant lambda cooling).

changed the description

I forgot to mention, but now the energy conservation seems to be better in the last graph, but we should wait on your fix of the cooling dependencies in order to be sure (the cooling is applied at the same time than the forces in this figure).

These only affected your section of the code so I am happy to merge them in.

Just to confirm:

• you do want to remove the wrapper file, right?
• you don't want to keep grackle2 ?

I do not see the point of having a wrapper if we are only calling grackle at one place (and in a small function). I have moved everything to cooling.h and cleaned the code.

By keeping grackle2, we will just have a piece of code that will rot. In this commit, I have done enough test to ensure that the implementation of grackle 3 gives the same answer than grackle 2, therefore there is no reason to keep grackle 2 as a benchmark

Ok, perfect. Just making sure, files were not disappearing by mistake.

Do you want to keed the branch?

merged

mentioned in commit c92464e2