typing.Namedtuple are collections.namedtuple in disguise. We do quite a lot of
work around understanding namedtuples, but we weren't doing the same for typing.NamedTuple.
In order to get a consistent experience, we moved typing.NamedTuple inference support
in the brain file corresponding to collections's namedtuple, and there we can leverage the
namedtuple inference rewriting functions. Now a new capability that typing.NamedTuple has is
to understand that we can define methods on the class, so that they can be used on the instance
when instantiated.
Another small bug that was fixed was to put Name nodes in .bases when building a namedtuple,
instead of putting the class node itself.

Close PyCQA/pylint#1628

This reverts commit 2df7ce47. cached was already used
in scoped_nodes, but my search didn't find it.

Makes load_module_from_file able to import files within the path that
are just a symlink to a file outside of the path.
modutils is using realpath at the moment which will resolve the symlink
and don't allow the import as the file is detected outside of the
available paths. By checking both abspath and realpath this patches add
support for such scenarios.

Get rid of the astpeephole (it's not an actual peephole and it's optimized just for once single use case

Close PyCQA/pylint#2055

52e67656 introduced a new type of
child iteration.
Some classes started yielding children in the incorrect order,
which was causing test failures in pylint.
This change corrects the ordering to match the _astroid_fields
of the class.

Relates to a change from #497

Some nodes such as the ExceptHandler has a name object but that is an AST node,
not a string. This fix prevents a crash when trying to repr() an ExceptHandler.

Because we don't support Python 2 any longer in the master branch, we can
return values from generators to signal that we want to throw a StopIteration,
without actually raising the StopIteration itself.

Super objects in python have a __qualname__ attribute
so super nodes should have an equivalent qname method.

Prevents an error in upstream pylint

Close #533

Probably the result of 3192f786

This allows inference of function arguments to be different than the
inference of the function body

Currently this is only used for object.__new__(cls) calls but
can be expanded later

The typing module is not available on Python3.4

It used to be a string containing the path, but it doesn't reflect the situation
on Python, where it is actually a list.
Also fix a bug with namespace package's __path__ attribute, which wasn't using the
module's __path__, but its file attribute, which is None for namespace packages.

Close #528

`_multi_line_block_fields` is a list of strings indicating which
fields contain multi-line blocks. `_get_multi_line_blocks` dynamically
accesses these attributes the first time it is called and then caches
the resulting references so as to avoid repeated expensive attribute
lookups.

Yield statements can only appear in multiline bodies (like function
definitions) and in lambdas, so there is no need to check other nodes
for them.

Return statements can only appear in multiline bodies (like function
definitions), so there is no need to check other nodes for them.

Assign statements can only appear in multiline bodies (like function
definitions) and in the value side of assign statements (e.g. `b = 4`
is an assign node contained in `a = b = 4`), so there is no need to
check other nodes for them.

This allows .root().name calls to work for Unknown objects

Close #523