Take all str/unicode methods, such as .capitalize(), .split(), .join(),
etc, and implement them for bstr/ustr. For example bstr.split() behaves
like unicode.split(), but returns list of bstr instead of list of
unicode. And similarly for all other methods.

Organize testing of this via verifying every method behaviour on all
unicode and bstr/ustr. If the results match by modulo of deep replacing
unicode to bstr/ustr - everything is ok.