It is posited that a resumed cluster may still wait up to the usual 2h+/-jitter until the next time it applies syncsets. Verify that this is true; and make it not so.

NB: I think it's acceptable if it's 2h+/-jitter after the last apply, if the resume happens sooner than that. E.g. the following:

• t=0 apply syncsets
• t=10m hibernate
• t=1h resume
• t=2h apply syncsets