* Historically, triple buffering was used to workaround missed VSYNC
  in GPU composition usecases on devices with high resolution panels.
  As a downside, an additional frame buffer increased input latency.
  Nowadays, most GPUs are fast enough to fit the VSYNC period, rendering
  triple framebuffers useless.
* Custom VSYNC offsets are intended to reduce input latency at cost of
  increased sync error counts. Our values were never calculated properly
  and were just taken from another device instead. That led to jank in
  some usecases.

Change-Id: I57ac4362ca28a9c128cc45c3f6e3734a24654955