fix(Hardware Support): Fix Dropped Events From Debounce Method

[pastaq]

In some rare circumstances the Legion Go S DPad can have some chatter when pressed, emitting a 'down', 'up', and 'down' event in the same event frame. The deck and deck-uhid targets both contain debounce sequences for double events, preventing a "no-op" if a 'down' then 'up' event is hit in the same frame by rescheduling the 'up' event for later, but it doesn't account for what should happen if another 'down' event is hit while the 'up' event is scheduled. In such a case, the second 'down' event becomes a no-op as the state doesn't change, then the deferred 'up' event hits, clearing the state and putting the virtual device out of sync with the physical hardware.

Additionally, all of the currently implemented target device scheduled_events() functions fail to check for the elapsed time has passed, instead immediately draining all scheduled events in the next poll.

This patch addresses these issues by making the following changes:

• Mask the Legion Go S evdev events to reduce the overall processing overhead for duplicated events and lower the chance for chatter.
• Dequeue pending 'up' events if a 'down' event is processed before that event's scheduled interval.
• Reduce ScheduledNativeEvent timeout on deck and deck-uhid to 8ms (approximately 2 frames) to reduce chatter collision probability.
• Check ScheduledNativeEvent.is_ready() for queued events before draining them on all target devices with implemented scheduled_events() functions.

Fixes ValveSoftware/SteamOS#2539