Perf improvement using zero-alloc foreach

[reyang]

#1078 (comment)

Potential places that we want to improve:

1. Processor chain (CompositeActivityProcessor).
2. Exporter interface (a batch of spans).
3. Resource attributes.

[reyang]

@CodeBlanch has an idea of using union approach to address the IEnumerator improvement among Processor/CircularBuffer/Exporter.

[martinjt]

There's been a fix related to this, if there's more work I'd suggest a more targeted suggestion.

[martinjt]

Most of the work has completed, if there's more work then you can reopen with the specific suggestions and why, or create a new targeted ticket for them.

I think there's been a tonne of work around allocations in the project, so it's probably something that has already been done. So I'm going to mark this as complete.