- Have the server clear the accept_stream callback prior to destroying
  the channel (required a small transport op protocol change)
- Have the transport not enact transport ops until parsing is completed
  (prevents accept_stream from disappearing mid-parse)

We suffered a bug whereby doing a follow-up write to another write could
resurrect a deleted stream, causing all sorts of crash.

Fix: when a stream becomes writable (vs when we start writing) take a
ref on the stream, and only relinquish it once we're done writing.

Some __dealloc__ methods were calling Python methods, and some
references were being dropped on the floor instead of threaded through
gRPC core.

That was overdue and the handshake is now slightly different for clients
(channels) and servers.

So multiple implementations can exist in one binary