Reduce bound checks in `take` kernels

**Is your feature request related to a problem or challenge? Please describe what you are trying to do.**
Take kernels perform bound checks for the kernels, as indices come from outside, however this is at the cost of some performance (~25% I remember from some earlier tests).

In the case of multiple arrays (batches) we perform the bound checks for each array, which is redundant.

**Describe the solution you'd like**

We can run consider having both bound-checked and non-bound-checked versions of the private kernels, and use them for `take_record_batch` for doing the `take` on `>1` arrays.

**Describe alternatives you've considered**

**Additional context**

Start at take_record_batch and trace the private take kernels it uses for single and multiple arrays. Compare where bound checks occur in each path; done means batches avoid repeating redundant checks while preserving the existing take behavior and performance goal.

Written by the indexing model from the issue text.