Aligns Connection.recv(timeout=0) with docstring

[mattyoungberg]

Fix #1552

NOTE: I added a couple of tests, but one of the two is still not passing. My intention is to fix and then squash the fix into this commit. At which point, this note will be removed.

The docstring for Connection.recv suggests that if you provide the argument timeout=0, you can retrieve any message in waiting:

"Set timeout to 0 to check if a message was already received."

Current behavior is that this does nothing and instead makes a TimeoutError immediately throw on the first use of a derived Deadline.

However, being able to check for a full message without blocking unneccesarily (aka, waiting for the server to send a message) would be useful, especially if your websocket client needs to be able to continue to do work in another thread and so periodically ceding control is important.

Most of the problem comes from the fact that the code doesn't treat a timeout value of 0 as special, and the code path will create a Deadline with it, and then rely on the side effects of blocking network I/O and exceptions in order to guarantee the contract of Connection.recv() which is to ultimately deliver a full message.

The error has been addressed by allowing the API of Assembler.get() to potentially return None when a new param, peek, is set to True (the default is False and maintains previous behavior). The intention of this param is to allow the method to acquire the class's mutex, check if there are any frames in the Assembler instance, and if not, return None. The only place where this param is used is within Connection.recv(), which is now able to treat the exceptional case of timeout=0 by setting up it's arguments to Assembler.get() without otherwise changing its code path.

Tests:

• Two tests were added: one to make sure that a client gets back None when timeout=0 and the underlying frames queue is empty, and another that returns the correct message when the underlying frames queue has a frame.

[mattyoungberg]

Closing in favor of #1573