SD: Bug fix for concentrated mass with CoG offset

[luwang00]

This PR should be ready to merge after review.

Feature or improvement description
This PR aims to fix a known bug with non-zero CoG offset/eccentricity of concentrated masses in SubDyn (Issue #1710) that was causing problems for some users (Issue #2451). The following fixes are implemented:

• The missing moments from the weight of offset concentrated masses are added during initialization when computing p%FG following the suggested fix of @ebranlard in Issue SD: concentrated mass with excentricity cannot be fully accounted for in SubDyn #1710.
• The node index, weight, and CoG offset of all concentrated masses are now saved in p%CMassNode. p%CMassWeight, and p%CMassOffset to support GuyanLoadCorrection for fixed-bottom structures and self-weight update for floating structures during the run.
• For fixed-bottom structures, an additional GuyanLoadCorrection term to the nodal moments is added. This term is in the form of cross_product( matmul(p%CMassOffset(i,:),orientation)-p%CMassOffset(i,:), (/0.0,0.0,-p%CMassWeight(i)/) ), where orientation is just the nodal orientation matrix due to the Guyan modes only. This additional correction term is zero if the offset/eccentricity is zero as expected.
• For floating structures, the additional force and moment from the weight of the concentrated masses are simply recomputed at each time step based on the Guyan (rigid-body) displacement of the structure. This follows naturally from the changes made in PR Support for large platform yaw offset in OpenFAST #2203.

This PR also adds a warning that SubDyn currently does not account for geometric stiffness from cable pretension if the user specifies one or more cables with non-zero pretension. See Issue #1956 and PR #2363.

Related issue, if one exists
Issues #1710, #2451, #1956

Impacted areas of the software
SubDyn

Test results, if applicable

• There is no change to existing test results. All r-tests passed.
• An additional test was performed with a fixed-bottom structure having a concentrated mass with a large CoG offset. The results after the bug fix are effectively identical to those generated with the workaround proposed in Issue SD: concentrated mass with excentricity cannot be fully accounted for in SubDyn #1710, i.e., using a rigid link to connect the node for attachment to the concentrated mass CoG to avoid having non-zero eccentricity.
• A similar test with a floating structure also shows consistent results between the two approaches.

[andrew-platt]

@RBergua, can you run a test of this?

[jjonkman]

I agree with the changes being implemented. Please merge once you are happy with the tests.

[RBergua]

I took advantage of the system documented here: #2021, defined two lumped masses at (0, 0, 0) and defined two eccentricities. I'm able to get the same results as using the rigid link + lumped mass workaround. So, everything looks alright to me.

Maybe I would clarify in the documentation that when defining an eccentricity, the moments of inertia that the user provides are around the center of mass.