[docker] Update Trilinos and refactor install process

[maxfirmbach]

Description and Context

As the title states.

Related Issues and Pull Requests

Closes #968 #675 #661 #192

[maxfirmbach]

I updated the wrong Trilinos hash ... before I fix that I'll just see if everything passes our test pipelines.

[maxfirmbach]

Locally it works for me, it seems like in the docker image SuperLU_dist does weird things ... not sure if the versions match.

[maxfirmbach]

@eulovi the following tests are failing, due to tolerances not reached:

embeddedmesh_parallel_bending_beam_fine.4C.yaml-p3 (Failed)
embeddedmesh_2d_hertzian_contact.yaml-p2-csv_comparison-xxx_energy.csv (Failed)
embeddedmesh_parallel_bending_beam_fine-vtk-structure.4C.yaml-p3 (Not Run)
embeddedmesh_block_nurbs27_hex8_wedge6-vtk.4C.yaml-p2 (Failed)

Would you mind looking at the results and check if they are acceptable? And if yes fix the relevant tests?

[maxfirmbach]

Seems like this PR triggers a really nice random behavior related to contact resulting in the failing of single tests on each pipeline run, just as usual.

[mayrmt]

I am fine with the current changes.

[maxfirmbach]

@ischeider @sebproell Any idea why the documentation pipeline fails?

[amgebauer]

@ischeider @sebproell Any idea why the documentation pipeline fails?

we treat warnings as errors in the documentation pipeline. There is the following warning in the pipeline:

/__w/4C/4C/build/doc/tmp/installation/installation.rst:102: WARNING: Inline strong start-string without end-string. [docutils]

[maxfirmbach]

@amgebauer I've just seen there's a build option to build with the documentation, will try that out locally! Thanks also for the hint about warning as errors.

[maxfirmbach]

Leaving this in Draft-PR mode for now, there are a few things that seem to go wrong (tests randomly failing).

[maxfirmbach]

Trilinos 16.2 will be the last version with Epetra. I wait for it's release, update this and make this the only allowed version to use with 4C. I will revert some things related to the install process here again.