Implement mj_stackAlloc and mj_arenaAlloc alignment using bitwise-and.

This implementation explicitly avoids calling division instructions (which are generally expensive) even when link-time optimization is not active or if the alloc functions aren't inlined for some reason.

Also make mj_stackAlloc cheaper for external callers who may be linking against MuJoCo dynamically (and therefore cannot benefit from inlining).

PiperOrigin-RevId: 561299565
Change-Id: Iea4be5bf6d351ddf0c9d1fd8bd89a0c207b39309

src/engine/engine_io.h

@@ -102,10 +102,10 @@ MJAPI void mj_resetDataDebug(const mjModel* m, mjData* d, unsigned char debug_va
 // reset data, set fields from specified keyframe
 MJAPI void mj_resetDataKeyframe(const mjModel* m, mjData* d, int key);
 
-// mjData arena allocate
+// mjData arena allocate (alignment must be a power of 2)
 MJAPI void* mj_arenaAlloc(mjData* d, size_t bytes, size_t alignment);
 
-// mjData stack allocate
+// mjData stack allocate (alignment must be a power of 2)
 MJAPI void* mj_stackAlloc(mjData* d, size_t bytes, size_t alignment);
 
 // mjData stack allocate for array of mjtNums