The current swap metrics average are hard to interpret because they are averaged over the entire session restore/background tab opening sessions. The duration of these sessions is highly variable, and the averaging will eliminate peaks.

The existing metrics should be deprecated and replaced with either one or both of:
1) Periodic metrics, sampled frequently (once per second) during sessions
2) Peak metrics, also sampled frequently but only emit the max rate during a session.

Prototype CL:
https://chromium-review.googlesource.com/c/chromium/src/+/646572