Optimizations and way better errors

• In a big effort, @paolobarbolini and me improved the performance of decoding sequences. This also lead to a cool contribution of him to the standard lib making vecdequeue for copyable types way more performant. #18
• I improved the reversed bitreader performance by a lot after reading the excellent blogposts here and here
• @rrbutani Made the effort of converting the not very helpful plain string errors with meaningful error enums, which apparently can help with binary sizes, but is also just a very big usability improvement of this library #28