Prevent yielding duplicated nodes from iterator

[akindyakov]

The problem of duplicates from node iterator comes from the fact that iterator is used as part of state being mutable object - every call to next() changes index in it. And in current version of SearchGrid iterator in state is never set explicitly, this is where most of the problems originates.

But there are more, another issue is that fetching callback can be called not exclusively. Meaning that there were multiple calls to it at the same time, which results in data race. I tried to solve this with throttling, but that didn't work very well. So I used a proper mutex now, from async-mutext package. Works like a charm.

Uploading Screen Recording 2023-02-12 at 16.53.19.mov…

Fixes #428

#351