Fixes test-encoding for python2.3 and minor cleanups:
- Use koi8-r instead of euc-jp, as the later isn't available in python2.3.
- Add unicode representation for latin1 e', too.
- Use date "1000000 0" like in other tests (though not strictly needed
because we no longer use %c as default datestr format).
35fb62a3a673d5322f6274a44ba6456e5e4b3b37 0 iD8DBQBEYmO2ywK+sNU5EO8RAnaYAKCO7x15xUn5mnhqWNXqk/ehlhRt2QCfRDfY0LrUq2q4oK/KypuJYPHgq1A=
2be3001847cb18a23c403439d9e7d0ace30804e9 0 iD8DBQBExUbjywK+sNU5EO8RAhzxAKCtyHAQUzcTSZTqlfJ0by6vhREwWQCghaQFHfkfN0l9/40EowNhuMOKnJk=