TimerResource replays every missed interval after an idle redraw

[zcourts]

Problem

Runtime::tick_resource_timers dispatches one action for every interval missed since next_fire_at:

while now >= *next_fire_at {
    ticks.push((action.clone(), payload.clone()));
    *next_fire_at = next_fire_at.saturating_add(interval_ms);
}

On the Web shell, the runtime clock advances by the elapsed wall time at the next redraw. If a tab has no redraw for 150 seconds, a two-second TimerResource dispatches approximately 75 actions at once on the next user interaction. A timer used for bounded network repair therefore becomes a request burst.

The same behavior is present in the 0.11.0 release and the current 0.11.1 release source.

Minimal reproduction

1. Declare a TimerResource with a two-second interval and an action that increments a counter or starts a job.
2. Let the Web app remain idle without redraws for approximately 30 seconds.
3. Trigger one redraw/user interaction.
4. Observe approximately 15 timer actions dispatched in that single tick.

The core behavior can also be reproduced deterministically by advancing a runtime clock by a multiple of the timer interval in one Runtime::tick call.

Expected behavior

A UI interval timer should coalesce missed intervals by default: dispatch at most one action when the runtime resumes, then schedule the next firing from the current runtime time. If catch-up replay is required for simulation use cases, it should be an explicit policy with a bounded maximum.

A focused test should cover a large clock advance and assert one coalesced dispatch for the normal timer policy.