Support for large platform yaw offset in OpenFAST

[luwang00]

Feature or improvement description

• The PR adds support for unrestricted platform rotation in ElastoDyn and SubDyn (for floating only).
• HydroDyn is currently limited to large yaw motion only with small roll and pitch offset.
• This PR also updates the general checking and handling of wave headings in the WAMIT and WAMIT2 modules for robustness. It should address some known issues when the range of wave headings crosses the +/-180-deg cut.
• The option to disable GuyanLoadCorrection in SubDyn is removed. GuyanLoadCorrection will always be used from now on.

Impacted areas of the software
Main impacted modules: ElastoDyn, HydroDyn, SubDyn
Minor updates for linearization: AeroDyn, MoorDyn, ServoDyn

Related Issue
For floating, the Guyan (rigid-body) modes of SubDyn models are made nonlinear and exact, consistent with ElastoDyn-only models. This change should address the discrepancies observed in Issue #2021.

Test results, if applicable

• r-test HydroDyn files are updated with the new inputs.
• With ElastoDyn and SubDyn switching to fully nonlinear platform rotation instead of small rotation approximation, the results of many existing r-tests involving a floating structure also changed and need to be updated.
• Some r-test results for the SubDyn module also changed because the option to disable GuyanLoadCorrection is removed, and all r-test results for SubDyn need to be regenerated with GuyanLoadCorrection set to true.
• HydroDyn r-test input files need to be updated because the input PRP motion file is now assumed to contain the time derivatives of the Tait-Bryan angles instead of the angular velocity and angular acceleration. HD driver will convert the input to angular velocity and angular acceleration internally.
• HydroDyn r-test output files also changed because some nonlinearity with rotational offset is introduced into the hydrostatic loads of potential-flow bodies.
• Several new r-tests with large yaw offset are added for the HydroDyn driver.
  See PR #133 to r-test.

Representative changes to r-test results are shown below:

5MW_OC3Mnpl_Linear
The eigenfrequencies from this test case show slight changes due to the changes made to the platform roll, pitch, and yaw DoF in ElastoDyn.

|             New            |     Reference (dev with     |
|                            |     GuyanLoadCorrection)    |
|  Freq (Hz)  |   Damp (%)   |   Freq (Hz)  |   Damp (%)   | 
|-------------|--------------|--------------|--------------|
| 0.017796474 | -0.777738405 | 0.017796428  | -0.777738379 |
| 0.017882340 |  0.779210600 | 0.017882306  |  0.779210674 |
| 0.275399247 |  0.001722274 | 0.275335134  |  0.001721133 |
| 0.277751794 |  0.001549770 | 0.277687353  |  0.001548639 |
| 0.712753964 |  0.004198908 | 0.712751224  |  0.004198887 |
| 0.743403938 |  0.004485754 | 0.743401949  |  0.004485744 |
| 0.995640417 |  0.005045851 | 0.995640050  |  0.005045819 |
| 1.091265363 |  0.005146327 | 1.091264164  |  0.005146313 |
| 1.191872529 |  0.005402909 | 1.191872058  |  0.005402910 |
| 1.654263201 |  0.019777407 | 1.654238948  |  0.019776211 |

5MW_OC3Spar_DLL_WTurb_WavesIrr
Note that the roll motion and, by extension, sway motion changed because ED now solves the nonlinear rotation in the order of intrinsic yaw->pitch->roll. While yaw is fairly small in this case, the moderate pitch displacement affects the roll response of the spar.

5MW_OC4Semi_WSt_WavesWN

[luwang00]

@andrew-platt I updated the MoorDyn and HydroDyn c-binding by replacing the calls to SmllRotTrans with EulerConstructZYX. I hope this is all we needed to do. It is the user's responsibility to make sure the data passed to HydroDyn and MoorDyn follow the adopted rotation convention and that the angular velocity and acceleration are computed correctly for large rotation. (Note that I have moved the yaw filter from ElastoDyn back into HydroDyn as discussed, so HydroDyn no longer requires any new inputs from the coupling code.)