Fix jacobian dimensionality alignment in ChainedTransform

The existing code only considered mixing of scalar and vector transforms, but not potentially higher dimensionality transforms (e.g., matrix)

Author: ricardoV94

pymc/logprob/transforms.py

@@ -895,7 +895,8 @@ def log_jac_det(self, value, *inputs):
         det = 0.0
         for det_ in det_list:
             if det_.ndim > ndim0:
-                det += det_.sum(axis=-1)
+                ndim_diff = det_.ndim - ndim0
+                det += det_.sum(axis=tuple(range(-ndim_diff, 0)))
             else:
                 det += det_
         return det