[FIX] l10n_at,l10n_tr: remove windows UTF8 BOM in Python source files, redundant with magic coding comment

Actually makes the babel term extractor fail with this
error:
  SyntaxError: python refuses to compile code with both
  a UTF8 byte-order-mark and a magic encoding comment

bzr revid: odo@openerp.com-20121123234256-2kcjfedy1czxd86l