Device log throttling issues and improvements

[gabrielburnworth]

• Frontend should receive throttled_at and throttled_until changes without a refresh. This is related to deviceIsThrottled and BatchQueue, which may not be working correctly.
• It should be possible to initiate throttling from any log type. Currently it takes a log type other than fun or debug to initiate the cooldown period when the rate limit has already been reached by those log types.
• Throttling should end when the cooldown period ends rather than after a log of a type other than fun or debug has been sent.
• Users should be notified upon termination of throttling (cooldown period) rather than after a new log message has been sent.
• Logs that would not display on any filter setting should be removed from frontend resources after the maximum log count has been reached.

To reproduce

1. Enable firmware response logs (firmware_input_log) with a device connected.
2. Wait a couple minutes. (Observe no throttling.)
3. Perform an action that creates a normal log near the end of a clock minute. (Observe "throttle begin" notice while logs continue to appear in frontend.)
4. Perform an action that creates a normal log near the beginning of the next clock minute. (Observe "throttle end" notice.)
5. Repeat, this time refreshing after step 3. (Observe logs no longer appear in frontend.)

[RickCarlino]

Will investigate item 1. Might be possible to send fatal_email when cooldown ends. The others probably won't get fixed, given current staffing.