Agents can wait for another task to finish, not just for a set time

Until now, an agent that depended on another task could only snooze until a set time, then check whether the other task was done. Often it wasn't, or the agent woke up long after it could have carried on. Now an agent can snooze until another card reaches a state, such as done, stopped or archived, and it wakes up as soon as that happens. When a snoozed card comes back, its log now says why: the other task is done, the time has come, you brought it back, there's something new for the agent (a review, a landing, a failed check), or the card needs you. Before, the log only said who had snoozed it.