Veselin Dobrev

Hi @giraldeau, is this PR ready for review from your perspective? If so, can you please remove the draft status and add the `ready-for-review` label? Thank you for your contribution!

@balos1 and I also talked about this -- we'll probably need to address it for the next xSDK release at the latest.

We should document (in the doxygen docs for `GetGeometricFactors` and `GetFaceGeometricFactors`) the need to call `Mesh::DeleteGeometricFactors` or the new `Mesh::NodesUpdated` when modifying the mesh in a way that will change...

Instead of filtering the modified big files, I think it will be better to compare them to the "merge base" version of the file. If the number of lines added...

There seems to be some misunderstanding regarding the use of `BilinearForm::InnerProduct()`, e.g. there is no bug in ex10p as it is in `master`. The inputs to `BilinearForm::InnerProduct()` are assumed to...

Looking briefly at this, I see some discussions that are not resolved yet, so I'll need to look closer. @psocratis, @jakubcerveny correct me if I'm wrong.

There's another PR with a similar goal: #1656.

@sebastiangrimberg, I see that you force-pushed this branch. Once review has started we ask people to NOT force-push to make it simpler for reviewers to follow changes in the PR...

There's another issue we ran into in our nightly testing: when building with the Intel compiler (our tests use version 2021.6.0 but other version probably have the same problem), SuperLU_DIST...

> I would think the simplest approach you suggest is to just copy the Hypre allocated data into new arrays which are SuperLU_DIST allocated at the expense of an extra...