Consider updating rotation matrices with Euler rotation

[SSL32081]

          Isn't these three rotation an euler rotation? If so, I think it is better to combine them into one rotation matrix instead of separating them, a single matrix operation should be much cheaper

Originally posted by @kazewong in #180 (comment)

TODO:

• Need to check the rotations matches in some corner cases
• Check accuracy difference