Fix phoenix http

[Jake-Shadle]

This fixes the issue where the phoenix HTTP server would stop responding after some amount of time. The issue was that since a single thread was being used for the service, it would deadlock itself accessing the Inner::nodes map, as the background_update_task could be context switched during an await while a write lock was held, causing the "main" loop to deadlock adding/retrieving nodes (or vice versa). I'm fairly confident this is the fix as now locks will eventually be released as tokio will be driving all tasks, not only a single one.

Resolves: #918

This also adds thread names to all of the threads and tokio workers spawned by quilkin for easier debug inspection. This also made me realize that because we use tokio::main we spawn a ton of extra workers in the proxy case, but I'll leave fixing that for a different PR.

[quilkin-bot]

Build Succeeded 🥳

Build Id: 8bb752ca-8d61-41bb-b9f9-64d1299af94d

The following development images have been built, and will exist for the next 30 days:

• us-docker.pkg.dev/quilkin/ci/quilkin:0.9.0-dev-2aab380

To build this version:

git fetch git@github.com:googleforgames/quilkin.git pull/926/head:pr_926 && git checkout pr_926
cargo build