Add micro-dumux-surrogate participant to two-scale heat conduction

[IshaanDesai]

This PR adds a surrogate variant to the existing participant micro-dumux, called the micro-dumux-surrogate. The surrogate is created using the tool BayesValidRox. The surrogate is built using PCE (Polynomial Chaos Expansion). This participant is built to highlight the functionality of model adaptivity in the Micro Manager: precice/micro-manager#198

Checklist:

• I added a summary of any user-facing changes (compared to the last release) in the changelog-entries/<PRnumber>.md.
• I will remember to squash-and-merge, providing a useful summary of the changes of this PR.

[MakisH]

I have not yet tried to run it, but here are some first, easy-to-address comments.

[MakisH]

The README.md file of the tutorial also needs to document this alternative and what kind of surrogate it is.

In other cases (e.g., perpendicular-flap), we had participants named -fake? Is this case doing something similar?

[MakisH]

What is this directory, and how is it related to the micro-dumux-surrogate/?
What does mada stand for?
Why is there no run.sh and clean.sh, if this is a case directory?

[MakisH]

Suggested change

	Micro simulation Surrogate, requrie previous computation of surrogate model
	Micro simulation Surrogate, requires previous computation of surrogate model

?

[MakisH]

See #680 for some updates (or directly copy from other run.sh in the same tutorial).

[IshaanDesai]

Thanks for the comments. I am still thinking about how to structure the folders so that only the full-order model and the surrogate model are exposed to the user. By adding a surrogate model at the micro scale, we have a unique scenario in which the Micro Manager runs not one but two solvers simultaneously. Technically, the solvers are not participants of the coupling; the Micro Manager is. But having the solvers in individual folders makes sense because it highlights which physics is being used and which software is being run.

[uekerman]

Thanks for the comments. I am still thinking about how to structure the folders so that only the full-order model and the surrogate model are exposed to the user. By adding a surrogate model at the micro scale, we have a unique scenario in which the Micro Manager runs not one but two solvers simultaneously. Technically, the solvers are not participants of the coupling; the Micro Manager is. But having the solvers in individual folders makes sense because it highlights which physics is being used and which software is being run.

Makes sense to me. The only real alternative would be a subfolder micro or micro-manager and in there multiple folders with the different models.