Add outputs to FV discretizers

[ftalbrecht]

No description provided.

[codecov]

Codecov Report

Merging #1196 (0c1fce3) into main (0a076cd) will decrease coverage by 0.01%.
The diff coverage is 72.58%.

Impacted Files	Coverage Δ
src/pymor/discretizers/builtin/fv.py	67.36% <71.66%> (+0.50%)	⬆️
src/pymor/operators/constructions.py	89.10% <100.00%> (+0.01%)	⬆️
src/pymortests/demos.py	90.12% <0.00%> (-1.24%)	⬇️
src/pymortests/vectorarray.py	92.02% <0.00%> (+0.11%)	⬆️
src/pymor/discretizers/builtin/cg.py	92.07% <0.00%> (+0.17%)	⬆️
src/pymor/vectorarrays/numpy.py	85.20% <0.00%> (+0.23%)	⬆️
src/pymor/bindings/ngsolve.py	79.05% <0.00%> (+0.67%)	⬆️
src/pymor/tools/deprecated.py	90.90% <0.00%> (+4.54%)	⬆️

[sdrave]

LGTM. But there is still some test missing that uses the code, right?

[ftalbrecht]

LGTM. But there is still some test missing that uses the code, right?

Yes. We could adapt a demo to include a time-depdent boundary output term and also plot a relative output error in the reduction_error_analysis (which would also resolve @renefritze request in #1179). However, we would need a MOR + FV demo, and the only one I'm seeing there is burgers_ei, which does not (yet) include the error analysis.

Suggestions, @sdrave?

[sdrave]

Suggestions, @sdrave?

Maybe show the elliptic demo some love and and compute some output quantities? No need for a time-dependent problem, I'd say.

[ftalbrecht]

Suggestions, @sdrave?

Maybe show the elliptic demo some love and and compute some output quantities? No need for a time-dependent problem, I'd say.

I would not call it love, but I added unconditional computation of both kind of outputs to the demo.

[sdrave]

@ftalbrecht, shall we merge this PR before it goes stale?

[ftalbrecht]

No objections from my side! However, I do not have any time at all at the moment (at least until Easter), for reasons.