Provide 'flatten' and 'flatten_async' in Request

[arteymix]

This feature provide flatten and flatten_async in Request to avoid the redundant and error-prone request body flattening.

The correct way to accumulate an InputStream have many drawbacks:

• the stream must be spliced into a MemoryOutputStream
• the MemoyOutputStream must be explicitly closed before calling get_data
• the data buffer size must be set from the MemoryOutputStream
• a fixed-size buffer can be used if the content length is known

Flattening could also be used for parts in a multipart message (see #81)