Use external transport crate #366

[gijsvl]

The extracted transport crate can be found here: https://github.com/boltlabs-inc/transport

We should make sure it all still makes sense, including the level of abstraction.
Specifically, the feature allow_explicit_certificate_trust, does that make sense to also live inside the transport crate, I do believe so, but just checking. Also, the abstraction for ZkChannelAddress that now implements the trait Address, which is part of the transport crate.

[marsella]

This looks good to me. Waiting to approve until I have a chance to double-check the transport crate, but it seems like you have a good abstraction. My comments are almost all formatting.

[marsella]

Review of the transport crate:

As discussed on slack, we might want to add a callback to the serve_while function (e.g. a FnOnce parameter) in order to allow custom logging or other administrative work once the server is running. However, I'm not sure if this would fix the logging issue we're seeing in tests (where we're currently filtering out non-zeekoe logs) because the log itself would be called from transport -- you will have to check the log and see. Options if the filtering is still a problem:

• Add the transport crate to the log filter
• Return to the original solution of sleeping for 5-10 seconds and assuming the server gets set up in that time

It looks to me like there aren't major changes in the transport crate except maybe adding an Address trait. Is there anything else specific that you'd like me to look at?

[gijsvl]

With regard to the transport crate, indeed, not much changed, it's basically the same with the exception of the abstract Trait Address, as you've mentioned.

[marsella]

Thank you for the fixes. My only remaining recommendation is to update the rustfmt.toml with the imports_granularity specification - it looks like that file already supports unstable features, although I'm not sure which ones. Mukund may have mentioned something about line-wrapping documentation, but I think that is an issue for a different PR.

[gijsvl]

Before merging, we should merge boltlabs-inc/transport#1, and change the transport dependency back to "main" branch

[marsella]

Looks good, approving pending the transport fix.