Enable module storage for 0.05% of visitors w/storage-capable browsers
Module storage was added in If2ad2d80d but hidden behind a feature flag and
disabled by default. Before we enable it by default, we want to make sure that
it really does improve performance. To gauge the performance impact of module
storage, we want to run a controlled experiment that will randomly assign 0.1%
of visitors with eligible browsers to either an experimental or control group.
We will collect page load timing data from both groups, but enable module
storage only for the experimental group. The logging code will live in the
NavigationTiming extension, but the group assignment logic has to happen before
any modules are actually loaded, so it has to be in core. The exact time-frame
of the experiment has not yet been determined but it will be in the
neighborhood of one week, after which I will remove this code from core. Bug
56397 provides a means for tracking my commitment to removing this code.
The experiment was designed by me and Aaron Halfaker.
Bug: 56397
Change-Id: Id2835eca4469310d79d28a243d4b0d76aa59cd2a