* dynamic filter network

* clean imports and get_data()

* change architecture

* a few modifications

* add readme and some rename

Bugfix for StagingArea. It appears as though Tensorflow is not adding it in the old contrib location. Should be backwards compatible.

* Remove assumption that history is 4 frames in expreplay.

* fix linting

* fix bug in boilerplate, self.cost and enhance FakeData

* fix len_range -> max_range

the mnist-convnet.py examples was overloaded. This commit splits the
file into the most minimal mnist-convnet and mnist-tfslim. Further
it adds an example about the visualization of filters which
illustrates the usage of `VariableHolder`.

In addition, this also adds a boilerplate template for projects,
which might facilitate the process of starting a new project.