Keep newly observed sleep credit until the following poll because the supplied idle duration can predate that sleep. Extend both contributor regression tests across the sampling boundary and qualify the platform guarantee.