telemetry: introduce orb-telemetry crate and use it in orb-ui

[TheButlah]

we should in general be using journald for our logs, not stdout. Also, we should use stderr instead of stdout.

Every one of our crates sets up logging in a different way, but they all do the same thing. I wanted to introduce a single spot that lets us configure our telemetry. If it goes well in orb-ui, we can use it in all the rest too

[fouge]

I tested and the filtering doesn't seem to work as expected

I have RUST_LOG=debug in my environment, but the same applies if the default_directive uses LevelFilter::DEBUG:

• when filtering out info and above levels with journalctl, i still have debug logs (ie tracing::debug!())
• these disappear only when I filter out warning and above levels 🤔

(venv) (os-b5c3a10) worldcoin@id-faff46c4:/mnt/scratch$ jl -fu worldcoin-ui --priority=info
-- Logs begin at Tue 2023-11-21 21:10:22 UTC. --
Oct 23 08:43:17 localhost worldcoin-ui[2338]: UI event: {"BatteryIsCharging":{"is_charging":true}}
[...]

corresponding line

tracing::debug!("UI event: {}", serde_json::to_string(event)?.as_str());

[fouge]

should we prevent the panic? return an error instead?

[TheButlah]

I think the panic is preferred. This type of code is once-only initialization code. If we called it more than once we should abort.

Generally errors are more for things that actually are expected to happen and need to be robustly handled. But someone calling this twice is better immediately slapped on the wrist and told to change their code.

[TheButlah]

when filtering out info and above levels with journalctl, i still have debug logs

This is expected, journald has a different naming scheme than the tracing crate. See here. Info level messages in journald are debug level in tracing.

[fouge]

should we document this somewhere? 🤔

[TheButlah]

Its already documented by tracing_journald.