replace for with .fold()

[chris-ha458]

This was kinda hacked into #44 as an after thought, but since it seems that might not be a good idea at this point, I extracted the fold() changes.

Whether this is more idiomatic code is subjective. Since subjective aspects would come down to the preference of the maintainer, I won't go too much into it. I will go into objective parts.

• for isn't strictly necessary since each closure, except for the Counter access has no other side effects
• this construction would make it easier for functional extensions, such as rayon::into_par_iter() (Although certain concurrency or atomicity related changes would need to be made for this)

[coriolinus]

Agree that for isn't strictly necessary here, but without getting into the question of what is more idiomatic, I'd prefer the for construction over fold for this case.

It's not clear what you mean by saying that this makes functional extensions easier. It's easy to imagine a rayon feature which when enabled does impl rayon::iter::ParallelIExtend for Counter, but the ParalllelExtend implementation would never just defer to the FromIterator implementation. Can you sketch out more precisely you expect this to make future implementation easier?

[chris-ha458]

but the ParalllelExtend implementation would never just defer to the FromIterator implementation. Can you sketch out more precisely you expect this to make future implementation easier?

I thought I'd do a simple rayon par_into_iter(), but it seems like it wouldn't be easy or even possible with the current bounds.
Another impl with different relevant bounds satisfied would be necessary and I get that even if such a construction was used to extend it functionally, it likely not look like the current version.

I might ponder on this a bit before closing, but feel free to close it earlier if you don't feel like there is anything here.