Fix minor bugs

[Oscar-Pepper]

Sync should be much more finely tuned with this PR

fixes:

1. to fix tokio polling being CPU-starved. moves located tree construction to scanning and uses spawn_blocking to keep rayon tasks in a separate thread pool to the main tokio thread
2. fixes a bug where workers may be shutdown before scan is complete
3. moves the point where a worker is a set to be scanning to fix a bug where workers are assigned multiple tasks
4. fixes a mempool panic, more work to be done on mempool in following PR

[zancas]

I'm confused.. is there a test? How can we know whether the sync is tuned, without observations?

[Oscar-Pepper]

I'm confused.. is there a test? How can we know whether the sync is tuned, without observations?

we dont have the benchmarking framework yet to show the performance improvements in the form fo a test.
i will edit the PR description to include more information about the changes / fixes

[zancas]

Are these bugs really "minor"? It seems like at least some are critical even if the effort to fix them is low. I am concerned that this code should be reviewed carefully, and "minor" might give the PR the wrong "flavor".