Dev: Logging

[TylerBloom]

Currently, there is no logging happening in the v2 pre-release. This will make fixing bugs rather difficult.

Ideally, any panics or issues with Discord would be logged locally and, when possible, sent to the error log on the dev server. The after_command event handler should be helpful for this.

[TylerBloom]

Serenity puts its panic handler around the calls to before and after. So, if the command fails, after is never called. The best possible solution is likely to add a listener. The before and after methods can pass actions to the listener via a channel. If the listener finds an action that lives for more than some amount of time, say 5 seconds, we will assume there was a panic and log the context.

In order to construct a better crash context, every command could track its progress and send messages to the listener. The listener could then piece together a crash context from these messages.

[TylerBloom]

This approach will also provide us with the tools to track general performance metrics.

[TylerBloom]

The logging should also include telemetry. Having daily or weekly reports on this would be helpful.