Much more useful now. :)

But fix UBSan complaint.

There's nothing in hb_bytes_t that guarantees lexical ordering, and
ordering by length first is much faster.

Ouch!

Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=11327

Reported as https://github.com/noporpoise/sort_r/issues/7

This reverts commit db889c18.

Was resulting in ambigious overloads...

https://github.com/harfbuzz/harfbuzz/pull/1334

It assumes all names are encoded in UTF16-BE.  Other than that, and not
listing languages correctly, it's *supposed* to work.

From bungeman.

Fixes https://github.com/harfbuzz/harfbuzz/pull/1314

Simplifies transient object creation.

Fixes https://github.com/harfbuzz/harfbuzz/issues/1301

Will come in through <T> if desired.

COLR table has one function: return layers for a glyph, and we expose exactly
that, so should just wire it through.  Also use sub_array() for verifiable
safety.

Also, BaseGlyphRecord's null object is enough.  We don't need to special-case
the not-found.

Some clang versions define static_assert as a macro apparently, so we cannot
redefine it...

This reverts commit 94bfea0c.
This reverts commit 4e626278.

So we don't accidentally break it again.

I don't think I broke anything.  Fuzzers will let me know..