Use banded solver for implicit flex integration.

The flex interpolation stiffness matrix within the implicit/implicitfast solvers is now built and factorized in a banded format instead of a dense one. This involves:
-   Calculating the bandwidth based on the sparsity of the mass/damping matrix and the connectivity within flex cells.
-   Allocating and populating a banded matrix `H`.
-   Using `mju_cholFactorBand` and `mju_cholSolveBand` for factorization and solving.
This change improves performance for flexes with many DOFs but local coupling.

PiperOrigin-RevId: 901297952
Change-Id: I3efe06353d1903ea65ab30dc49685cede228bb68

include/mujoco/mjmodel.h

@@ -978,6 +978,7 @@ struct mjModel_ {
   int*      flex_matid;           // material id for rendering                (nflex x 1)
   int*      flex_group;           // group for visibility                     (nflex x 1)
   int*      flex_interp;          // interpolation (0: vertex, 1: nodes)      (nflex x 1)
+  int*      flex_bandwidth;       // precomputed solver bandwidth             (nflex x 1)
   int*      flex_cellnum;         // finite cell num per dimension            (nflex x 3)
   int*      flex_nodeadr;         // first node address                       (nflex x 1)
   int*      flex_nodenum;         // number of nodes                          (nflex x 1)

include/mujoco/mjxmacro.h

@@ -454,6 +454,7 @@
     X   ( int,     flex_matid,            nflex,         1                    ) \
     X   ( int,     flex_group,            nflex,         1                    ) \
     X   ( int,     flex_interp,           nflex,         1                    ) \
+    X   ( int,     flex_bandwidth,        nflex,         1                    ) \
     X   ( int,     flex_cellnum,          nflex,         3                    ) \
     X   ( int,     flex_nodeadr,          nflex,         1                    ) \
     X   ( int,     flex_nodenum,          nflex,         1                    ) \