Add parallel iteration for exact chunks

[Choudhry18]

Implement Rayon IntoParallelIterator support for ExactChunks and ExactChunksMut by wrapping their NdProducer traversal in an unindexed parallel producer. Add coverage for immutable and mutable exact chunk iteration and document the new API.

Closes #1192

[nilgoyette]

This is not a critic of your code (I haven't checked), but I wonder if we actually want a _par version of our functions. I mean, it will double our API. However, most (all?) iterators/producers can be used with Zip::par_*. Are you telling me thatExactChunks can't?

Sorry @akern40 to disturb you :) Do you have an opinion on this _par matter?

[Choudhry18]

ExactChunks can be used with Zip::par the comment on the issue also mentions that. My understanding was that we would want to add the _par versions for consistency as AxisIter and AxisChunksIter both have _par versions when they can also be used with Zip::par

[akern40]

Not a bother! Sorry for the slow response - I see both sides of the argument. I would rather we get rid of all the par helpers, I think? But for now, I think consistency continues to be paramount. I'm in favor of the _par methods, and we will need to figure out how to factor out a more minimal API in the future.