fix travis with Chainerv5

[yuyu2172]

tests/links_tests/model_tests/ssd_tests/test_multibox_loss.py:141: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
communicator_name = 'naive', mpi_comm = None, allreduce_grad_dtype = None
    def create_communicator(
            communicator_name='hierarchical', mpi_comm=None,
            allreduce_grad_dtype=None):
        """Create a ChainerMN communicator.
    
        Different communicators provide different approaches of communication, so
        they have different performance charasteristics. The default communicator
        ``hierarchical`` is expected to generally perform well on a variety of
        environments, so one need not to change communicators in most cases.
        However, choosing proper communicator may give better performance.
        The following communicators are available.
    
        +---------------+---+---+--------+--------------------------------------+
        |Name           |CPU|GPU|NCCL    |Recommended Use Cases                 |
        +===============+===+===+========+======================================+
        |pure_nccl      |   |OK |Required|``pure_nccl`` is recommended when     |
        |               |   |   |(>= v2) |NCCL2 is available in the environment.|
        +---------------+---+---+--------+--------------------------------------+
        |hierarchical   |   |OK |Required|Each node has a single NIC or HCA     |
        +---------------+---+---+--------+--------------------------------------+
        |two_dimensional|   |OK |Required|Each node has multiple NICs or HCAs   |
        +---------------+---+---+--------+--------------------------------------+
        |single_node    |   |OK |Required|Single node with multiple GPUs        |
        +---------------+---+---+--------+--------------------------------------+
        |flat           |   |OK |        |N/A                                   |
        +---------------+---+---+--------+--------------------------------------+
        |naive          |OK |OK |        |Testing on CPU mode                   |
        +---------------+---+---+--------+--------------------------------------+
    
        Args:
            communicator_name: The name of communicator (``naive``, ``flat``,
              ``hierarchical``, ``two_dimensional``, ``pure_nccl``, or
              ``single_node``)
            mpi_comm: MPI4py communicator
            allreduce_grad_dtype: Data type of gradient used in All-Reduce.
              If ``None``, the dtype of a model is used.
    
        Returns:
            ChainerMN communicator that implements methods defined in
            :class:`chainermn.CommunicatorBase`
    
        """
    
        if mpi_comm is None:
            try:
                import mpi4py.MPI
            except ImportError as e:
>               raise ImportError(str(e) + ": "
                                  "ChainerMN requires mpi4py for "
                                  "distributed training. "
                                  "Please read the Chainer official document "
                                  "and setup MPI and mpi4py.")
E               ImportError: No module named mpi4py.MPI: ChainerMN requires mpi4py for distributed training. Please read the Chainer official document and setup MPI and mpi4py.

[Hakuyume]

LGTM