The e-mail messages are normalized in a way compatible with both Python 2 and Python 3.

In Python 2, the previous version added a unixfrom, which contains the date, making the tests fail when there is a change of one second in the clock between generating the first and the second message. Moreover, the second message contained a previous unixfrom. This has been solved by removing unixfrom in the normalized string.

In Python 3, messages had an inconsistent folding, due to https://github.com/python/cpython/issues/72649 . This has been solved by manual unfolding (using a regex substitution) and by outputing normalized strings without folding (setting maxheaderlen=0, which means no maximum line lenght).

With these changes, the normalized string can be compared in both versions without problems.

In addition, some bare encode/decode calls were replaced by the appropriate methods.

Tests passed locally and a test suite is in progress.