idle_analysis: Fix idle residency when idle events are sparse

Created by: bjackman

Not too sure if this is the best approach.. maybe we should come up with a more generic solution (Trace._sanitizeCpuIdle?). I haven't investigated whether _dfg_cluster_idle_state_residency is affected by the same issue (in my brief tests it didn't seem to be but good chance that's coincidence).

This is kind of a similar issue to this: https://github.com/ARM-software/lisa/pull/220