Any objection to a PR to support bytearray in addition to array.array buffers? bytearray() has the advantage that you can specific the length when you create it instead of needing an initializer object.

Context is this commit which added the implicit readinto support: 1028dc6