Fix appveyor's older MSVC++ builds by working around 2015 Update 2 bugfix

MSVC++ before 2015 Update 2 has a bug in sleep_for where it tries to
implicitly += the input with a nanoseconds variable. Work around this by
using nanoseconds directly (which can be implicitly +='d with
chrono::nanoseconds).

Support for manual timing (see issue #198, https://github.com/google/benchmark/issues/198)

Only output optional fields if they're set

Close a memory leak in configuration code

Add section on iterations.

Also add some subheadings, and fix up the line lengths.

Fixes #194

Fix SetLabel usage example

Some benchmarks may run a few milliseconds which makes it kind of hard to visually compare, since the currently only available nanoseconds numbers can get very large in this case. Therefore this commit adds an optional command line flag --benchmark_time_unit which lets the user choose between ns and ms time units for displaying the mean execution time.

Fix missing include in map_test.cc

building with libc++.

Pass const State to Fixture::TearDown. Fix memory leak in fixture_test

Format json in README.md

Pass const state to Fixture::SetUp. Add map_test.

Add number of threads to State.

Having access to the thread count from within a benchmark is useful,
for when one wants to distribute a workload dynamically among the
benchmarks running in parallel e.g when using ThreadRange() or
ThreadPerCpu().

Add section on optimisations to README

Add aarch64 support to src/cycleclock.h.

Fix ARM /proc/cpuinfo parsing.

Support MSVC on appveyor

Also fix related warning message typos.

From google3's base/cycleclock.