Removes one more static-initialization.  A few more to go.

This was causing all object types to be non-POD and have static
initializers.  We don't need that!

Now, most nil objects just moved from .bss to .data.  Fixing for that
coming soon.

According to Tom Hacohen this was breaking build with some compilers.

In file included from hb-buffer-private.hh:35:0,
                 from hb-ot-map-private.hh:32,
                 from hb-ot-shape-private.hh:32,
                 from hb-ot-shape.cc:29:
hb-object-private.hh: In constructor '_hb_object_header_t::_hb_object_header_t()':
hb-object-private.hh:97:8: error: uninitialized const member in 'struct hb_reference_count_t'
hb-object-private.hh:51:25: note: 'hb_reference_count_t::ref_count' should be initialized
In file included from hb-ot-shape.cc:33:0:
hb-set-private.hh: In constructor '_hb_set_t::_hb_set_t()':
hb-set-private.hh:37:8: note: synthesized method '_hb_object_header_t::_hb_object_header_t()' first required here
hb-ot-shape.cc: In function 'void hb_ot_shape_glyphs_closure(hb_font_t*, hb_buffer_t*, const hb_feature_t*, unsigned int, hb_set_t*)':
hb-ot-shape.cc:521:12: note: synthesized method '_hb_set_t::_hb_set_t()' first required here

We never use them in fact...

I'm just adjusting these as I better understand the requirements of
the code and the guarantees of each operation.

Done.  I promise.

The OSAtomicAdd32Barrier operator returns the new value, we want the
old value.

So, apparently there's no atomic int 'get' method on Apple.  You have to
add(0) to get.  And that's not const-friendly.  So switch inert-object
checking to a non-atomic get.  This, however, is safe, and a negligible
performance boost too.

Apparently it can't be implemented on OS X.  We weren't using it anyway.

By defining HB_NO_MT.

Also, only warn once per missing MT feature support.

Mozilla Bug 666661 - gfx/harfbuzz/src/hb-prive.h - compiler warnings on mac

Reported by Jonathan Kew.

We need this to set data on objects safely without worrying that some
other thread unsets it by setting it at the same time.

Where causing issues for people with MSVC.

Looks *so* much nicer!

Buffer debugging coming soon.

Such that user_data and other finalizers cannot resurrect object

The test/object test is passing again, instead of deadlocking.

For now, by taking a global user_data mutex.

Mutex (and Windows.h by extension) are fairly isolated now.

The Win32 definitions for LONG, ULONG, etc conflicts with
hb-open-type.h.  Avoid that by making sure hb-object-private.h
and hb-open-type.h are not included in the same compilation unit.