Ouch!  Missing coords should still be evaluated as coord=0, which
most of the time results in a factor of 0.  We were skipping these,
which was equivalent to a factor of 1.

Fixes https://github.com/harfbuzz/harfbuzz/issues/652

Previously we made CGJ unskippable.  Now, if CGJ did NOT prevent
any reordering, allow skipping over it.  To make this work we
had to make changes to the Arabic mark reordering algorithm
implementation to renumber moved MCM marks.  See comments.

Fixes https://github.com/harfbuzz/harfbuzz/issues/554

In macOS 10.12, the `OSMemoryBarrier` and related APIs were deprecated
in favor of using `std::atomic`. On the way to supporting `std::atomic`,
we can favor using the "Intel primitives" which are also available on
macOS.

Due to how ninja runs clang and gcc, it doesn't colorize the output
like make does. This forces color output in this situation.

Fixes https://github.com/harfbuzz/harfbuzz/pull/661

Fixes https://github.com/harfbuzz/harfbuzz/pull/660

https://github.com/harfbuzz/harfbuzz/issues/542#issuecomment-353169729

Fixes https://github.com/harfbuzz/harfbuzz/issues/657

Otherwise setting the CTFont was ineffective as it would have been
recreated anyway unless font size was set to 18 CSS points.

Specificaly, when a range or sorted array has unexpected order, we take that as
font data being garbage and bail out.  This fixes significant slowdown on a bad
version of Chandas font which has a 600KB GPOS with garbage inside.

Later on, I like to add a maximum-work counter for collect_glyphs to protect
against malicious fonts as well.

Fixes https://bugs.chromium.org/p/chromium/issues/detail?id=794896

Does page lookup as needed.

The coverage already covered that.

The three "XXXXX"'s should be switched to false.  Doing that separately for ease
of bisecting...