Open
Description
As part of debugging #38269, we've been discussing how we can indicate in tests that completion failed because of a timeout instead of a genuine failure. We don't want to return an error in this case because we still want to return whatever completion items we were able to produce before the timeout. @findleyr suggested that we could use the telemetry library for this.
/cc @ianthehat