implement Read::read_to_string on top of Decoder::decode_to_string

This is a possibly somewhat common method, and the decoder in this crate could implement it more efficiently by avoiding UTF-8 re-validation. See @hsivonen's comment here: https://github.com/hsivonen/encoding_rs/issues/8#issuecomment-411356371

Locate the implementations of Read::read_to_string and Decoder::decode_to_string in this crate, then read the linked encoding_rs discussion for the intended efficiency constraint. Confirm how the adapter currently handles decoded text and use the existing test suite to verify that read_to_string works without unnecessary UTF-8 re-validation.

Written by the indexing model from the issue text.