Simplify Protobuf format

[grodtron]

Currently protobuf is wrapping all messages in the following global type, and holding the actual message as a serialized string, forcing a second manual serialization and deserializtion of the message itself.

message DebugMsg {

  enum Type {
    Run = 1;
    Next = 2;
    ...
  }

  required Type type = 1;
  required bytes message = 2; // protobuf serialized message
}

This (I believe) was done as a work around for the fact that the nanomsg python library attempts to decode data to be sent as ascii, and crashes on non-ascii characters (e.g. '\xff'). Some testing has shown that if we instead pass the serialized protobuf message as a list of individual characters, it will not attempt to do an ascii decode and will send the raw data correctly:

# In terminal 1
>>> import nnpy
>>> sock = nnpy.Socket(nnpy.AF_SP, nnpy.REQ)
>>> sock.connect("ipc:///tmp/nn_testing")
>>> s = '\xff\xaa\x00 This is a \xBB\xCC test'
>>> sock.send(s)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python2.7/dist-packages/nnpy/socket.py", line 53, in send
    data = data.encode() if isinstance(data, str) else data
UnicodeDecodeError: 'ascii' codec can't decode byte 0xff in position 0: ordinal not in range(128)
>>> sock.send([c for c in s])
>>> 

# In second terminal
>>> import nnpy
>>> sock = nnpy.Socket(nnpy.AF_SP, nnpy.REP)
>>> sock.bind("ipc:///tmp/nn_testing")
>>> sock.recv()
'\xff\xaa\x00 This is a \xbb\xcc test'
>>> 

This workaround I believe was also forcing other odd limitations such as using strings to represent boolean or integer fields.

A good amount of the debugger code could be simplified based on this, and would definitely be an improvement in the overall quality of the project.