Fast page-wise matrix decomposition

Perform commonly used matrix decomposition (QR, EIG, SVD) for N-D arrays using C-Mex and LAPACK library.

NOTE

recommend using the builtin `pagesvd` (R2021b), `pageeig` (R2023a).

When I was working on human motion analysis and inverse kinematics, SVD and EIG had to be called for many many many times.(for estimating rotations)

As the for-loop was quite slow, I tried to use mex-file and LAPACK library to speed up the performance.

There are three functions provided.

MDQR is page-wise QR for N-D arrays.

MDEIG is page-wise EIG for N-D arrays.

MDSVD is page-wise SVD for N-D arrays.(As MATLAB has introduced a new function called `pagesvd`, recommend using that for stability and performance. `MDSVD` might be deprecated.)

Before using these functions, you must compile the mex-file (directly call the function without any input argument to compile).

---------------------------------------------------------------------------------------------------

I don't know much about c-language, so the source code is ugly.

I would appreciate if you could kindly leave your comments.

Please be nice haha.