Pluggable request & response processing pipeline

[Andrey Zaytsev]

I'd be cool to have a pluggable data pipeline for cases when you want to:

1. pre-process responses, dumping "raw" text content in files by using common logger interface, for example;
2. pre-process requests, adding correlation ids in headers / adding checksums or custom signatures.

Additional layer required to make those things transparent to the client, without a need to tailor each particular request or create a special highly specialized wrapping "clients" for particular case.