Found and fixed a couple bugs.

Found a couple multithreading issues.  Marked them with "XXX-MT-bug".

This reverts commit 5daf3bd4.

If other atexit callbacks try to destruct the objects we destruct
in atexit callbacks, bad things will happen.

I'll come up with some other way to catch premature destruction
of HB-owned objects.

Fixes https://github.com/behdad/harfbuzz/issues/618

Such that client cannot accidentally destroy them, even though that
will be a bug in their code...

Oops.

The new unorm2_normalize has been public API since ICU 4.4, while
the old unorm_normalize has been deprecated since ICU 56.

Now, beginning with ICU 60, unorm_normalize will also be annotated
U_DEPRECATED and trigger the compiler -Wdeprecated-declarations warning.

Fix compile error in hb-icu.cc when ICU configured with
U_NO_DEFAULT_INCLUDE_UTF_HEADERS=1

As reported by Steven Loomis, including uversion.h works everywhere.

It's considerably faster than the fallback implementation we had
previously!

Based on patch from Philip Withnall.
https://bugs.freedesktop.org/show_bug.cgi?id=41095

The latter is what I meant.

Can be -1 for NUL-terminated string.  This is useful for passing parts
of a larger string to a function without having to copy or modify the
string first.

Affected functions:

	hb_tag_t hb_tag_from_string()
	hb_direction_from_string()
	hb_language_from_string()
	hb_script_from_string()

Where causing issues for people with MSVC.

Add tests.

Spacing_Mark is the current Unicode long-name for this property value.
The previous name was wrongly carried from glib.

Add compose() and decompose() unicode funcs.  These implement
pair-wise canonical composition/decomposition.

The glib/icu implementations are lacking for now.  We are adding
API for this to glib, but I cannot find any useful API in ICU.
May end of implementing these in-house.

Changed all unicode_funcs callback names to remove the "_get" part.
Eg, hb_unicode_get_script_func_t is now hb_unicode_script_func_t,
and hb_unicode_get_script() is hb_unicode_script() now.