Close #14074

We introduce a new package structure for Coq:

- `coq-core`: Coq's OCaml tools code and plugins
- `coq-stdlib`: Coq's stdlib [.vo files]
- `coq`: meta-package that pulls `coq-{core,stdlib}`

This has several advantages, in particular it allows to install Coq
without the stdlib which is useful in several scenarios, it also open
the door towards a versioning of the stdlib at the package level.

The main user-visible change is that Coq's ML development files now
live in `$lib/coq-core`, for compatibility in the regular build we
install a symlink and support both setups for a while.

Note that plugin developers and even `coq_makefile` should actually
rely on `ocamlfind` to locate Coq's OCaml libs as to be more robust.

There is a transient state where we actually look for both
`$coqlib/plugins` and `$coqlib/../coq-core/plugins` as to support
the non-ocamlfind plus custom variables.

This will be much improved once #13617 is merged (which requires this
PR first), then, we will introduce a `coq.boot` library so finally
`coqdep`, `coqchk`, etc... can share the same path setup code.

IMHO the plan should work fine.

On windows we provide a way to set environment variables
local to a coq installation by providing a file
named "coq_environment.txt" containing KEY="value" pairs.

No space between KEY and = is allowed, values are in
quotes according to OCaml's escaping conventions.
The file is line-directed, illformed lines are skipped.

This allows proper treatment in notations, ie fixes #13303

The "glob" representation of universes (what pretyping sees) contains
only fully interpreted (kernel) universes and unbound universe
ids (for non Strict Universe Declaration).

This means universes need to be understood at intern time, so intern
now has a new "universe binders" argument. We cannot avoid this due to
the following example:

~~~coq
Module Import M. Universe i. End M.
Definition foo@{i} := Type@{i}.
~~~

When interning `Type@{i}` we need to know that `i` is locally bound to
avoid interning it as `M.i`.

Extern has a symmetrical problem:

~~~coq
Module Import M. Universe i. End M.
Polymorphic Definition foo@{i} := Type@{M.i} -> Type@{i}.
Print foo. (* must not print Type@{i} -> Type@{i} *)
~~~
(Polymorphic as otherwise the local `i` will be called `foo.i`)

Therefore extern also takes a universe binders argument.

Note that the current implementation actually replaces local universes
with names at detype type. (Asymmetrical to pretyping which only gets
names in glob terms for dynamically declared univs, although it's
capable of understanding bound univs too)

As such extern only really needs the domain of the universe
binders (ie the set of bound universe ids), we just arbitrarily pass
the whole universe binders to avoid putting `Id.Map.domain` at every
entry point.

Note that if we want to change so that detyping does not name locally
bound univs we would need to pass the reverse universe binders (map
from levels to ids, contained in the ustate ie in the evar map) to
extern.

I left the other test as a v file since it might become relevant when the
corresponding Qed bug is fixed.