add kademlia tests

[orlandohohmeier]

Add Kademlia (integration) tests, improving the test coverage from 17.95 % (Lines), 35.46 % (Functions) to 82.37 % (Lines), 51.77 % (Functions). Some of the error cases aren't well covered yet.

In case you've never check coverage run the test with the instrument-coverage flag:

RUSTFLAGS="-C instrument-coverage" cargo test

And then generate a report using grcov as follows:

grcov . -s . --binary-path ./target/debug/ --branch --ignore-not-existing -t html

Side Note: Use cargo install --locked grcov) to install grcov if you haven't already.

This will give you a nice html report in ./html that you can open in your browser to understand the coverage.

This is the first PR in a series to improve test coverage pulled from #38

[codecov]

Codecov Report

❌ Patch coverage is 96.36364% with 16 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
crates/network/tests/kad_test.rs	96.36%	16 Missing ⚠️

📢 Thoughts on this report? Let us know!

[nfnt]

This looks good overall, but let's document why we need sleep in the testing code so that we can remove it in the future.

[nfnt]

LGTM!

[nfnt]

Nit, but a complicated one because of re-exports in multiple packages: IMO, we don't need the futures package at all, because we can find all of it either in futures_util or in tokio. futures re-exports these types from futures_util, the only difference is that it includes an async runtime similar to tokio but which we don't use. I.e. by using futures_util::stream::StreamExt instead and consistently, we can remove a dependency on futures.
tokio also provides its own version of StreamExt, but that ones different from the one in futures_util, missing a lot features.
Right now our code is using all of these in different places and IMO, we should at least remove futures in favor of futures_util, maybe not here but in a separate commit.