Use BufferedReader in get_readable_fileobj - #8634

@pllim - any reason for this to wait for 4.0?

I don't think there is a gain in doing this. The idea of using BufferedReader (as I interpret it from the comment) is to avoid reading the full buffer if it is not needed. But as soon as BytesIO is used with a .read(), the buffer is read and we can continue directly with BytesIO.
So at first I think that we need to find in which case the change would be useful, and then we can decide if the additional complexity introduced by #7981 is worth it or not.