fix: random walk

[jacobheun]

Random Walk had some bugs, this fixes them:

• If you ever specified more than 1 query, it would never run them. (You really shouldn;t need more than 1, but still not a good bug)
• We timed out the walk, but findPeer wasn't given the walks timeout. So if you specified 10seconds as your walk timeout, the findPeer logic would keep running until its default timeout kicked in (60 seconds)

I added a suite of tests just for RandomWalk to test the failures I was seeing.

This also adds support for AbortController in random walk and passes it to findPeer, however, findPeer currently isn't listening to the abort signal so it won't actually stop the query. My thought is to wait until we finish #82 and then add the abort logic to all of the query methods. Once we do, random walk will immediately benefit because it already passes the signal in. @dirkmc @vasco-santos let me know what you think, and I can update accordingly.

[dirkmc]

It's great to have this really comprehensive testing for random walk 👍

I'm not sure I'd include the AbortController logic until we move to an async/await API

[vasco-santos]

I am not against adding the AbortController logic in this PR. But we should have a comment explaining it will not be used for now / WIP

[jacobheun]

I've removed the runningHandler logic and simplified the code. I also bumped up the default timeout to 30s from 10s. Since we are passing on the timeout to findPeer, and it actually times the query out, giving it more time to run will improve the effectiveness.

I also disabled random walk in the TestDHT util by default to keep the tests isolated better. We're managing connects as needed in the tests, having random walk running will just cause more test chaos.

I added a comment to the RandomWalk._query function about the AbortController not yet being properly supported and removed the docs for options.signal from findPeer.

[jacobheun]

Now that timeout is actually being passed to findPeer I bumped this to 10seconds so the random walk doesn't get killed after 1s. This was causing test failures because routing tables weren't doing enough connecting when it was depending on the walk.

[vasco-santos]

LGTM

[dirkmc]

One comment but otherwise LGTM 👍

Thanks for cleaning up how the tests work as well!