Switch from env_logger crate to tracing for more in-depth observability info

[emanguy]

We're currently using env_logger for basic logging in the backend, but to really help us troubleshoot stuff in production we should migrate to the tracing crate. This has essentially the same API for basic logging, but it also enables us to use tracing spans and export trace data to external services such as Google Cloud Trace or AWS Distro. Doing this will help us troubleshoot problems in the backend and help identify performance issues.