sensors: Rewrite the frequency hint test.

This test has been flaky on the Mac bots for a very long time. Previous
efforts such as r770328 ('sensors: Make frequency hint test wait for another
slow sensor "tick"') did improve things a bit, but the flakiness has not
been fully fixed.

The problem is that our mock platform sensor dispatches sensor reading
notifications via window.setInterval(), and the actual period of the
callback can vary significantly, to the point where we cannot make any
assumptions about how precise it will be. The Mac bots in particular seem to
have big deviations in the timer's precision.

Consequently, we also cannot assume a faster sensor will really receive more
"reading" events than a slower one, or how long it will take for it to
happen.

Make the test deterministic at the expense of testing a less "real" use
case: instead of comparing how many times each sensor receives a "reading"
event, we just check that MockSensor.getSamplingFrequency() is adjusted
accordingly depending on which sensors are active.

Bug: 731018
Change-Id: Idc994d7e7d1cfb30d7c4e2f4b285494faff7160c
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2401139


Commit-Queue: Raphael Kubo da Costa <raphael.kubo.da.costa@intel.com>
Auto-Submit: Raphael Kubo da Costa <raphael.kubo.da.costa@intel.com>
Reviewed-by: Reilly Grant <reillyg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#805674}