Process subtrees of the schema that do not specify type

[danijar]

Already works with the default key but write tests to check if other types work as well. Solution might be to change the conditional from

def _parse(self, name, schema, definition):
    has_type = schema and 'type' in schema
    if definition is not None and not has_type:
        return definition
    # ...

to

def _parse(self, name, schema, definition):
    if definition is not None and not schema:
        return definition
    # ...