HTTP receiver should not expose bookkeeping functions

[lance]

In receiver_binary.js there is a check() function which is used to validate the input. It is currently defined on BinaryHTTPReceiver.prototype but I don't think it should be exposed at all. It's the SDK's job to check input and validate it, not the user of the SDK, so why expose it? I suspect it's useful for testing. In fact, that's how the tests are written at the moment. The same exists for the structured receiver.

Since the parse() function already calls check(), I think it's reasonable to remove this function from the prototype and make it internal to the module.

Are there any objections to this?