Review all no 'u'-prefixed string literals

[ramiro]

AFAICT at this point we have the following cases:

• String literals in modules that weren't touched to add from __future__ import unicode_literals (to fix Remove unicode() calls #25). In that state these literals are byte-buffers on Python2 and Unicode text on Python3. We need to make sure we avoid this behavior change possibly by expanding our use of unicode_literals.
• String literals in modules that were touched to fix switch to unicode_literals #16 by removing unicode(). e.g. 6382ee3#L3L198. This needs further fixing in presence of the fix for Remove unicode() calls #25.
• String literals in modules that were touched to add from __future__ import unicode_literals (to fix Remove unicode() calls #25). We need to make sure we prefix them with 'b' or leave them alone according to the original intent and context. I did this in 164c6e0 for two test-related files but I'd like to review all the code.

[ramiro]

Advances made in be5d805.