Describe the problem

The ioTokensExhaustedDurationMetric is busted on v22.1 because every time setAvailableIOTokensLocked is called when sg.availableIOTokens < 0, it re-adds now() - sg.exhaustedStart to the metric without resetting the value of sg.exhaustedStart. On a node that's constantly overloaded, it ends up with a ioTokensExhaustedDurationMetric that increases roughly quadratically, such that taking its rate gives a linearly increasing graph over time, which doesn't seem to be the metric's intent. See this where @a-robinson spotted it.

Jira issue: CRDB-19062

Epic CRDB-54646