Add latency and token per second stats to `LLMCallEvent`.

[ekzhu]

Confirmation

• I confirm that I am a maintainer and so can use this template. If I am not, I understand this issue will be closed and I will be asked to use a different template.

Issue body

It's useful to tracing LLM call performance.

Work on this once #5730 is completed so we can add this to all LLM related events.