The format strings that are known at compile time are now accessible
via both ::value and ::format(), and format strings for everything
else is accessible via ::format(). This makes it backwards compatible.

Add pybind11::make_key_iterator for map iteration

This allows exposing a dict-like interface to python code, allowing
iteration over keys via:

    for k in custommapping:
        ...

while still allowing iteration over pairs, so that you can also
implement 'dict.items()' functionality which returns a pair iterator,
allowing:

    for k, v in custommapping.items():
        ...

example-sequences-and-iterators is updated with a custom class providing
both types of iteration.

Improve constructor/destructor tracking

This commit rewrites the examples that look for constructor/destructor
calls to do so via static variable tracking rather than output parsing.

The added ConstructorStats class provides methods to keep track of
constructors and destructors, number of default/copy/move constructors,
and number of copy/move assignments.  It also provides a mechanism for
storing values (e.g. for value construction), and then allows all of
this to be checked at the end of a test by getting the statistics for a
C++ (or python mapping) class.

By not relying on the precise pattern of constructions/destructions,
but rather simply ensuring that every construction is matched with a
destruction on the same object, we ensure that everything that gets
created also gets destroyed as expected.

This replaces all of the various "std::cout << whatever" code in
constructors/destructors with
`print_created(this)`/`print_destroyed(this)`/etc. functions which
provide similar output, but now has a unified format across the
different examples, including a new ### prefix that makes mixed example
output and lifecycle events easier to distinguish.

With this change, relaxed mode is no longer needed, which enables
testing for proper destruction under MSVC, and under any other compiler
that generates code calling extra constructors, or optimizes away any
constructors.  GCC/clang are used as the baseline for move
constructors; the tests are adapted to allow more move constructors to
be evoked (but other types are constructors much have matching counts).

This commit also disables output buffering of tests, as the buffering
sometimes results in C++ output ending up in the middle of python
output (or vice versa), depending on the OS/python version.

Silence MSVC warning

PR #329 generates the following warning under MSVC:

    ...\cast.h(202): warning C4456: declaration of 'it' hides previous local declaration

This renames the second iterator to silence it.

Track registered instances that share a pointer address

reference_internal requires an `instance` field to track the returned
reference's parent, but that's just a duplication of what
keep_alive<0,1> does, so use a keep alive to do this to eliminate the
duplication.

It no longer applies since instances are now identified by both address
and type.

The pointer to the first member of a class instance is the same as the
pointer to instance itself; pybind11 has some workarounds for this to
not track registered instances that have a registered parent with the
same address.  This doesn't work everywhere, however: issue #328 is a
failure of this for a mutator operator which resolves its argument to
the parent rather than the child, as is needed in #328.

This commit resolves the issue (and restores tracking of same-address
instances) by changing registered_instances from an unordered_map to an
unordered_multimap that allows duplicate instances for the same pointer
to be recorded, then resolves these differences by checking the type of
each matched instance when looking up an instance.  (A
unordered_multimap seems cleaner for this than a unordered_map<list> or
similar because, the vast majority of the time, the instance will be
unique).

Move support for return values of called Python functions

Currently pybind11 always translates values returned by Python functions
invoked from C++ code by copying, even when moving is feasible--and,
more importantly, even when moving is required.

The first, and relatively minor, concern is that moving may be
considerably more efficient for some types.  The second problem,
however, is more serious: there's currently no way python code can
return a non-copyable type to C++ code.

I ran into this while trying to add a PYBIND11_OVERLOAD of a virtual
method that returns just such a type: it simply fails to compile because
this:

    overload = ...
    overload(args).template cast<ret_type>();

involves a copy: overload(args) returns an object instance, and the
invoked object::cast() loads the returned value, then returns a copy of
the loaded value.

We can, however, safely move that returned value *if* the object has the
only reference to it (i.e. if ref_count() == 1) and the object is
itself temporary (i.e. if it's an rvalue).

This commit does that by adding an rvalue-qualified object::cast()
method that allows the returned value to be move-constructed out of the
stored instance when feasible.

This basically comes down to three cases:

- For objects that are movable but not copyable, we always try the move,
  with a runtime exception raised if this would involve moving a value
  with multiple references.
- When the type is both movable and non-trivially copyable, the move
  happens only if the invoked object has a ref_count of 1, otherwise the
  object is copied.  (Trivially copyable types are excluded from this
  case because they are typically just collections of primitive types,
  which can be copied just as easily as they can be moved.)
- Non-movable and trivially copy constructible objects are simply
  copied.

This also adds examples to example-virtual-functions that shows both a
non-copyable object and a movable/copyable object in action: the former
raises an exception if returned while holding a reference, the latter
invokes a move constructor if unreferenced, or a copy constructor if
referenced.

Basically this allows code such as:

    class MyClass(Pybind11Class):
        def somemethod(self, whatever):
            mt = MovableType(whatever)
            # ...
            return mt

which allows the MovableType instance to be returned to the C++ code
via its move constructor.

Of course if you attempt to violate this by doing something like:

    self.value = MovableType(whatever)
    return self.value

you get an exception--but right now, the pybind11-side of that code
won't compile at all.

Fixed finding python libraries on windows in venv