Timeout should measure from last activation rather than first activation.

[amacleod]

After setting timeout-ms to 1500, I have observed that the third behavior gets activated 1.5 seconds after the first behavior, regardless of how many times I activate the middle behavior.

The way I want it to work, which I think is consistent with the description in the readme, is for it to activate the third behavior 1500 milliseconds after the last time I touch the key (or keep the timer alive by touching an ignored key).

Setting timeout-ms will cause the deactivation behavior to fire when the time has elapsed after releasing the Tri-State or a ignored key.

[amacleod]

Practically speaking, I probably do not need timeout-ms, and can instead rely on the interrupt functionality. The only downside I can imagine is for my alt-tab tri-state to just leave alt pressed for a long time until I decide to press something else.