[performance] Touch scrolling is stuttery under Unity8 (but only while you're touching it)

Mako also benefits a bit:

restart unity8-dash MIR_CLIENT_PERF_REPORT=log
tail -f ~/.cache/upstart/unity8-dash.log

[1439963004.335110] perf: Scopes: 53.83 FPS, render time 16.29ms, buffer lag 39.79ms (3 buffers)
[1439963005.338590] perf: Scopes: 54.83 FPS, render time 16.02ms, buffer lag 38.80ms (3 buffers)
[1439963006.340451] perf: Scopes: 55.94 FPS, render time 16.14ms, buffer lag 36.86ms (3 buffers)
[1439963007.351133] perf: Scopes: 47.52 FPS, render time 19.31ms, buffer lag 43.84ms (3 buffers)
[1439963008.363494] perf: Scopes: 57.31 FPS, render time 15.04ms, buffer lag 37.34ms (3 buffers)
[1439963009.384706] perf: Scopes: 56.80 FPS, render time 16.13ms, buffer lag 36.59ms (3 buffers)
[1439963010.451729] perf: Scopes: 49.67 FPS, render time 18.56ms, buffer lag 39.77ms (3 buffers)
[1439963011.457985] perf: Scopes: 58.64 FPS, render time 15.30ms, buffer lag 38.11ms (3 buffers)
[1439963012.468362] perf: Scopes: 56.43 FPS, render time 16.90ms, buffer lag 36.48ms (3 buffers)

restart unity8 QML_NO_TOUCH_COMPRESSION=1
restart unity8-dash MIR_CLIENT_PERF_REPORT=log
tail -f ~/.cache/upstart/unity8-dash.log

[1439963121.853194] perf: Scopes: 61.87 FPS, render time 12.15ms, buffer lag 36.61ms (3 buffers)
[1439963122.863754] perf: Scopes: 59.40 FPS, render time 10.45ms, buffer lag 40.20ms (3 buffers)
[1439963123.866348] perf: Scopes: 59.88 FPS, render time 14.06ms, buffer lag 36.28ms (3 buffers)
[1439963124.875901] perf: Scopes: 60.45 FPS, render time 10.91ms, buffer lag 38.34ms (3 buffers)
[1439963125.930898] perf: Scopes: 55.97 FPS, render time 12.45ms, buffer lag 39.29ms (3 buffers)
[1439963126.976923] perf: Scopes: 59.27 FPS, render time 12.67ms, buffer lag 38.09ms (3 buffers)
[1439963128.001339] perf: Scopes: 60.54 FPS, render time 11.60ms, buffer lag 39.43ms (3 buffers)
[1439963129.009977] perf: Scopes: 61.50 FPS, render time 12.01ms, buffer lag 36.53ms (3 buffers)
[1439963130.014616] perf: Scopes: 55.77 FPS, render time 14.85ms, buffer lag 39.11ms (3 buffers)

Status changed to 'Confirmed' because the bug affects multiple users.

Yes, using just Mir's 59Hz input rate by itself would appear to solve this bug. However looking ahead we need a different solution. Because we'll soon be ready to re-enable Mir's dynamic queue scaling that can provide significantly lower latency. However for touch-driven apps it only works when the input rate is at least that of the display. So we need Mir's input rate raised to 60Hz, or to fix the Qt touch compression and use that instead (non-nested, so only in unity8 but not apps).

Considering also related bug 1497105 and bug 1488386, I think all three would benefit from us removing input resampling completely from the top-level (unity8 process) via:
   QML_NO_TOUCH_COMPRESSION=1
   MIR_CLIENT_INPUT_RATE=0

That would immediately solve this bug, and bug 1497105, and partially solve the keep-awake problem of bug 1488386.

The only issue then is to make sure shell elements are still responsive in the presence of high frequency input events. Apps themselves would not be affected because they already have their own resampling.

Annoyingly, in order to reliably set QML_NO_TOUCH_COMPRESSION, you have to do it in the parent process before unity8 or the Qt client is launched. Because it's only checked once on static initialization:

static bool qquickwindow_no_touch_compression = qEnvironmentVariableIsSet("QML_NO_TOUCH_COMPRESSION");

So whatever is launching unity8 (upstart job?) has to set QML_NO_TOUCH_COMPRESSION. And the same goes for Qt apps.