Changes needed to run 300_pointneurons example with PyNN

[apdavison]

This is not for merging immediately, as these changes make it incompatible with the current development version of bmtk.

The main changes are:

• put "target_simulator" in "circuit_config.json" instead of "simulation_config.json" (this is also more compatible with the current version of the spec)
• add a prefix "nest:" to "static_synapse" in the "edge_types.csv" files. This is not absolutely essential, but makes the edge_types files more consistent with the node_types files.
• specify the population name in "node_sets.json" - this I think is required by the spec.

[apdavison]

With these changes, "run_PyNN.py" gives identical results to "run_pointnet.py" (the latter run using the current master branch).

[apdavison]

• one more change I forgot to mention. PyNN is putting the data in "/report/internal/data" in the HDF5 file, which is how I interpreted the spec, while BMTK puts it in "/data". I modified "plot_potential.py" to accommodate both locations.

[apdavison]

Run with PyNN r28861a5239 ("serialization" branch)