Feature: Adding minmax, similar to how numpy and jax recently added. #4569

☑️ I understand it is strictly prohibited to use AI to write issues.

So Recently numpy added a new method np.minmax which returns both min and max at the same time numpy/numpy#32231, It performs better than individual min and max operations in there benchmarks, numpy/numpy#32231 (comment), I am not sure if the performance improvement will translate to the MLX itself, but I can benchmark and see.

Jax has also recently added it jax-ml/jax#40089, I think it could be nice addition if it improves the performance.

If you guys think it could be a nice addition, i can work on this :)