Remove Epetra

[mperego]

This PR removes all the code related to Epetra and all the tests associated to Epetra. Before the removal, the Epetra tests that did not have a Tpetra analog have been converted to Tpetra tests.

The build_type has been removed, so "Build Type" needs to be removed from input yaml files.

[mcarlson801]

It looks like some of the tests are failing with this branch on weaver:

 46 - landIce_FO_GIS_Unstructured_Restart (Failed)
 61 - landIce_FO_GIS_SensSMBwrtBeta_Restart (Failed)
 64 - landIce_FO_GIS_Humboldt_LaplacianPriorSampling (Failed)
 99 - corePDEs_SideSetLaplacian_3D (Failed)

The FO_GIS tests are failing to find decomposed meshes:

Exodus Library Warning/Error: [ex__check_file_type]
    ERROR: Could not open file '/home/maxcarl/weaver/builds/albany-remove_epetra/tests/landIce/FO_GIS/gis_unstruct.exo.4.3', error = No such file or directory.

I think this is something Luca just fixed for other tests. Do we need to add something to the inputs for these tests?

The SideSetLaplacian_3D test is failing with the following error:

99: ************************************************************************
99: -- Nonlinear Solver Step 49 --
99: ||F|| = 2.500e-01  step = 1.000e+00  dx = 0.000e+00
99: ************************************************************************
99:
99: Ifpack2::ILUT::Compute: zero pivot encountered! Replacing with rownorm and continuing...(You may need to set the parameter 'fact: absolute threshold'.)
99: Ifpack2::ILUT::Compute: zero pivot encountered! Replacing with rownorm and continuing...(You may need to set the parameter 'fact: absolute threshold'.)
99: Ifpack2::ILUT::Compute: zero pivot encountered! Replacing with rownorm and continuing...(You may need to set the parameter 'fact: absolute threshold'.)
99: Ifpack2::ILUT::Compute: zero pivot encountered! Replacing with rownorm and continuing...(You may need to set the parameter 'fact: absolute threshold'.)
99:
99:   *******************************************************
99:   ***** Belos Iterative Solver:  Block Gmres
99:   ***** Maximum Iterations: 100
99:   ***** Block Size: 1
99:   ***** Residual Test:
99:   *****   Test 1 : Belos::StatusTestImpResNorm<>: (2-Norm Res Vec) / (2-Norm Res0), tol = 0.0001
99:   *******************************************************
99:   Iter   0, [ 1] :    1.000000e+00
99:   Belos::BlockGmresSolMgr::solve(): Warning! NaN has been detected!
99: WARNING: NOX::Direction::Newton::compute() - Linear solve failed to achieve convergence - using the step anyway since "Rescue Bad Newton Solve" is true
99: NOX::Solver::LineSearchBased::iterate - using recovery step for line search

Not sure why this would fail but not the 2D version.

[mperego]

It looks like some of the tests are failing with this branch on weaver:

 46 - landIce_FO_GIS_Unstructured_Restart (Failed)
 61 - landIce_FO_GIS_SensSMBwrtBeta_Restart (Failed)
 64 - landIce_FO_GIS_Humboldt_LaplacianPriorSampling (Failed)
 99 - corePDEs_SideSetLaplacian_3D (Failed)

The FO_GIS tests are failing to find decomposed meshes:

Exodus Library Warning/Error: [ex__check_file_type]
    ERROR: Could not open file '/home/maxcarl/weaver/builds/albany-remove_epetra/tests/landIce/FO_GIS/gis_unstruct.exo.4.3', error = No such file or directory.

I think this is something Luca just fixed for other tests. Do we need to add something to the inputs for these tests?

The SideSetLaplacian_3D test is failing with the following error:

99: ************************************************************************
99: -- Nonlinear Solver Step 49 --
99: ||F|| = 2.500e-01  step = 1.000e+00  dx = 0.000e+00
99: ************************************************************************
99:
99: Ifpack2::ILUT::Compute: zero pivot encountered! Replacing with rownorm and continuing...(You may need to set the parameter 'fact: absolute threshold'.)
99: Ifpack2::ILUT::Compute: zero pivot encountered! Replacing with rownorm and continuing...(You may need to set the parameter 'fact: absolute threshold'.)
99: Ifpack2::ILUT::Compute: zero pivot encountered! Replacing with rownorm and continuing...(You may need to set the parameter 'fact: absolute threshold'.)
99: Ifpack2::ILUT::Compute: zero pivot encountered! Replacing with rownorm and continuing...(You may need to set the parameter 'fact: absolute threshold'.)
99:
99:   *******************************************************
99:   ***** Belos Iterative Solver:  Block Gmres
99:   ***** Maximum Iterations: 100
99:   ***** Block Size: 1
99:   ***** Residual Test:
99:   *****   Test 1 : Belos::StatusTestImpResNorm<>: (2-Norm Res Vec) / (2-Norm Res0), tol = 0.0001
99:   *******************************************************
99:   Iter   0, [ 1] :    1.000000e+00
99:   Belos::BlockGmresSolMgr::solve(): Warning! NaN has been detected!
99: WARNING: NOX::Direction::Newton::compute() - Linear solve failed to achieve convergence - using the step anyway since "Rescue Bad Newton Solve" is true
99: NOX::Solver::LineSearchBased::iterate - using recovery step for line search

Not sure why this would fail but not the 2D version.

I'll fix the tests failing for the error with loading the mesh. As for the SideSetLaplacian_3D test, it must be some uninitialized variable or something like that. On my workstation it converges in 1 iteration. I'll see if I can find the culprit, but I might go ahead and push the PR anyway, and address the error later.

[mperego]

OK, I have renamed the tests to remove references to Tpetra and hopefully fixed the exudus issues in the test files. I could not reproduce the issue with SideSetLaplacian_3D on weaver. But I'd rather push the PR and we look at that later.

[mperego]

Thanks @bartgol and @mcarlson801 for your feedback!