Speed up coordinate_vector function

[user202729]

Build up on #40497, coordinate_vector can use the same optimization.

the v = v.list() is an unfortunate artifact of another suboptimality, namely we have the following

    cpdef _coerce_map_from_(self, M):
        """
        Return a coercion map from `M` to ``self``, or None.

        The implementation of this method in module classes should
        observe the following guidelines:

        1. We want to relate two different *ambient* modules if and
           only if they have the same rank (which is the same as
           degree) and if there is a coercion of the base rings.

        2. Two modules embedded in other modules that have a coercion
           may inherit a coercion if they are in fact sub-modules of
           one another.

        3. Since different embeddings of one abstract module are
           related by non-identical coerce maps (in 2.), we must not
           have coercion from a sub-module to the corresponding
           abstract module, for otherwise non-commuting coercion
           triangles emerge.

        4. Quotient modules must not coerce unless their modulus `W`
           is the same. There must not be forgetful maps.

        5. Coerce maps for quotient modules are already registered on
           construction.

        TESTS:

        Make sure :issue:`3638` is fixed::

            sage: vector(ZZ,[1,2,11]) == vector(Zmod(8),[1,2,3])                        # needs sage.modules
            True

        AUTHORS:

        - Simon King (2010-12)

        - Peter Bruin (June 2014)
        """
        try:
            if (isinstance(M, Module)
                    and self.base_ring().has_coerce_map_from(M.base_ring())
                    and M.change_ring(self.base_ring()).is_submodule(self)):
                return M.hom([self._element_constructor_(x, check=False) for x in M.gens()], self)
        except (TypeError, NotImplementedError, AttributeError, ArithmeticError):
            pass
        return None

this map is called even to convert a from a submodule of ℤ^2 to ℤ^2.

the problem is ModuleMorphism needs to (1) call coordinate_vector, then (2) multiply it by the basis matrix. and (1) leads to an infinite loop.

Requires #40371

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes.
• I have updated the documentation and checked the documentation preview.

⌛ Dependencies

#40371

[github-actions]

Documentation preview for this PR (built with commit a3ab0d3; changes) is ready! 🎉
This preview will update shortly after each push to this PR.