[discretizers.cg] affine decomposition of robin operator and rhs functionals

[TiKeil]

This PR addresses #727. For now the discretizer does not decompose the rhs functional F and the diffusion part of the robin boundary data although LincombFunctions might be given.

In this PR, the discretizer check for all inputs whether the input is a LincombFunction. If that's the case, the rhs functionals and robin operators in the model will also be decomposed now

[codecov]

Codecov Report

Merging #760 into master will decrease coverage by 0.03%.
The diff coverage is 61.29%.

Impacted Files	Coverage Δ
src/pymor/discretizers/cg.py	83.89% <61.29%> (-6.43%)	⬇️

[codecov]

Codecov Report

Merging #760 into master will decrease coverage by 0.03%.
The diff coverage is 61.29%.

Impacted Files	Coverage Δ
src/pymor/discretizers/cg.py	83.89% <61.29%> (-6.43%)	⬇️

[sdrave]

Thanks for your work! You should add an entry to AUTHORS.md. Maybe you could add some minimal tests for the new code, e.g. by modifying a demo script?

[TiKeil]

I changed AUTHORS.md.

To modify a demo script is a little bit less trivial than I thought since the elliptic.py is not parametric at all and elliptic2 has no neumann or robin conditions. Where would it fit best then?

I propose that we first merge this and discuss editing the demo later?