Performance improvements for take

**Is your feature request related to a problem or challenge? Please describe what you are trying to do.**
Take is a critical function that is used in quite some code and is hot up in profiling. It seems like we should be able to speed it up.

Two improvements we can do:
* Bound checks are about 25% of the work (in take benchmark)
* SIMD could be utilized.

**Describe the solution you'd like**
Check whether we can eliminate / avoid bound checks.
Add SIMD versions of take

**Describe alternatives you've considered**

**Additional context**