Match.pos from Pypy's RE engine inconsistent from CPython
In CPython, in a pattern.finditer() context, all matches' pos are assigned to the original startpos of finditer(pos=...).
While in Pypy, pos is updated thus increased in every internal search().
print([match.pos for match in re.finditer("a", "aaa")])
CPython output: [0, 0, 0]
Pypy output: [0, 1, 2]