Automatically chunk data sent to Stream.WriteData

[BrianBland]

Stream.WriteData always constructs exactly one spdy DataFrame, which causes an error when the data byte array has length greater than 16MB (the spdy max frame data size).

I'm proposing that when large (>16MB) data is sent to WriteData, we should automatically split this into multiple data frames. This change would cause a disconnect between WriteData and ReadData as currently implemented, as ReadData always expects a single data frame.