Document best practices for high performance constructing arrays

[alamb]

          > I think at this point there is little point in using `MutableBuffer` over `Vec` as the latter provides more performant (specialization over `T`, better inlining), slightly more safe and more complete API. The same probably applies for a lot of `Builder`-type APIs probably.

It might be worth spending some time documenting the best and most performant way to construct / transform Arrow arrays and apply it accross the arrow-rs create (marking stuff as deprecated if needed).

This LGTM and is a great improvement. I also would love to see this documentation if you don't mind dumping your thoughts on the topic while they're fresh!

Originally posted by @mbutrovich in #7422 (review)

[Dandandan]

Let me start with collecting a list of possible replacements of MutableArray vs Vec-based construction and faster generally faster way of doing things, so we could convert it

Vec-based API

Old	Vec-based
MutableBuffer::new(size * std::mem::size_of::<T>())	Vec<T>::with_capacity(size)
BufferBuilder	Vec
Buffer::from_trusted_len_iter	Iterator::collect (into Vec) - this doesn't use unsafe and is as fast
Primitive builders	Use Vec<T> (Iterator::collect) and build nulls separately (via BooleanBuffer::collect_bool if possible)
Byte builder	Use Vec<T>, Vec<Offset> (Iterator::collect) and build nulls separately (via BooleanBuffer::collect_bool if possible)

Faster versions

Slower	Faster
MutableBuffer::push or Vec::push	extend or collect
vec![] or `MutableBuffer::new(0)	Vec::with_capacity (preallocate once for known capacity)
NullBufferBuilder or boolean buffer builder	BooleanBuffer::collect_bool

[alamb]

Related:

• Consider using Vec directly in builders #7383