Use ifaces instead of objs, stop wrapping everything in two (or three)
layers of no-value-added indirection, and remove some of the more
pointless/outdated idioms from the code.
having a read_chars method is convenient and more efficient.
the old read_char method had a bug due to re-use of the 'w' width
variable as a loop counter and so was broken for wide characters,
this patch fixes that.
This parallels nicely with string_reader and string_writer, but
working with raw byte strings. I'm not quite sure if the return
type for bytes_writer.get_bytes should be [u8] or [mutable u8]
though, so I'm arbitrarily picking mutable for now.
