* faster and more informative training plots

* Update utils.py

Looks good. Needs pep8 linting, I'll do that in PyCharm later once PR is in.

* Update test.py

* Update train.py

f for the tb descriptor lets us plot several batches, i.e. to allow us to change L292 to 'if ni < 3' for 3 examples.
Co-authored-by: NGlenn Jocher <glenn.jocher@ultralytics.com>

As per https://github.com/ultralytics/yolov3/issues/1096

In a clean environment running training fails if tensorboard is not installed e.g.

```
Traceback (most recent call last):
  File "/home/josh/miniconda3/envs/ultralytics/lib/python3.7/site-packages/torch/utils/tensorboard/__init__.py", line 2, in <module>
    from tensorboard.summary.writer.record_writer import RecordWriter  # noqa F401
ModuleNotFoundError: No module named 'tensorboard'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "train.py", line 6, in <module>
    from torch.utils.tensorboard import SummaryWriter
  File "/home/josh/miniconda3/envs/ultralytics/lib/python3.7/site-packages/torch/utils/tensorboard/__init__.py", line 4, in <module>
    raise ImportError('TensorBoard logging requires TensorBoard with Python summary writer installed. '
ImportError: TensorBoard logging requires TensorBoard with Python summary writer installed. This should be available in 1.14 or above.
```