v0.1.2

Optimization to reduce syscalls using a buffered reader. Increases throughput by ~12% on reading from high-bandwidth/volume channels (e.g. file transfers), at a tiny cost on memory (4KiB buffer per conn).