The iter objects shouldn't not be const.  D'oh.

The bool conversion happens after predicate is called automatically.

Allows for eg, checking all values equal 2: hb_all (it, 2).

They do absolutely same thing.  hb_requires is to encode constraints,
whereas hb_enable_if is for more conditional enabling.

Part of https://github.com/harfbuzz/harfbuzz/issues/1648

The first three based on range-v3.

Fixes https://github.com/harfbuzz/harfbuzz/issues/1651

This reverts commit 98f14c4c.

Same as previous commit.

Painful.  All template methods need to be explicitly hidden :(.

Maybe we should switch to -fvisibility=hidden pragma.

A LOT more to go.

Don't think we need hb_is_same().

From Orvid King: TLDR; MSVC has some issues using sizeof(declval<T>()) for
SFINAE of templated types, so I just used SFINAE in a different context where
MSVC doesn't have the issue.

While doable with hb_deref_pointer() as well, we also would then
need to do it in a ton of places.  Not worth it / messy.

No idea how to avoid dupicating code.  Was hoping hb_deref_pointer()
would do it, but looks like a pointer can't bind to a universal
reference T&&.  Humm.

Yeah, some things not very clear...

Untested.