CFD coupling to use AD15 mesh only

[andrew-platt]

This PR is ready for merging

Feature or improvement description
The OpenFOAM module had been using both the ElastoDyn structural mesh (for loads) and the AeroDyn 15 mesh (for motions). After much in person discussions with @tonyinme, @gantech, @michaelasprague, @jjonkman, and @ebranlard
(see #1232), we settled on using only the AeroDyn 15 mesh.

This also allows BeamDyn to be used with CFD codes.

Related issue, if one exists

• First step of corrections in Incorrect force mesh used with OpenFOAM module #1232
• Closes Force points and velocity points are in different locations in ALM Exawind/amr-wind#673

Impacted areas of the software
Glue code:

• Pass tower and blade info from AD15 instead of ElastoDyn

OpenFOAM module:

• remove all ElastoDyn information
• remove unnecessary intermediate meshes in mapping (extra Line2 mesh)
• Some code factoring (indentation mostly)
• improved error handling

Additional supporting information
The original problem was a mismatch in location of motion points and force points passed to CFD. The velocity points were taken from the AD15 input mesh (interpolated to finer discretization), but the force points were set based on the ElastoDyn blade mesh (also refined discretization).

This PR is a short term fix for the mesh issue. The long term fix will involve modifications to AD15 and interface changes for SOWFA, NALU-Wind, and AMR-Wind:

• AD15 interpolate points to refine the mesh
  • use nearest neighbor for polar
• will require passing to CFD the point locations and number of points instead of receiving the number of points

Recommend squash merge for cleaner history.

Test results, if applicable
@tonyinme will post relevant info from AMR-Wind tests inline below

[jjonkman]

I have not reviewed the source code changes (and don't intend to), but I agree with the solution as discussed.

[ebranlard]

I think the code is much cleaner this way.

I suggest to add more details in the description of the datatypes (just so that it's easier to get back to it in the future).

Optionally, maybe we can use InterpArray to shorten the code and make it more readable.

[andrew-platt]

Some updates from @tonyinme:

Overlay of the force and velocity measurement locations. The force positions are shown for 3 timesteps and the velocity positions are shown for one timestep. The latest timestep for force position overlaps with the velocity locations as expected.

[andrew-platt]

Comparison of tangential forces before and after the fixes in this PR (green is prior to fixes, red is with fixes):

[andrew-platt]

For anyone following this PR, I squashed down the commit history for a cleaner looking merge commit.