50% of regex_v8 benchmark time is 2 regexes

[dg-pb]

These 2 lines in block1:

        regexs[17].sub(r'', strings[2], count=subcount[17])

        # This prints a unicode escape where the V8 version prints the
        # unicode character.
        regexs[17].sub(r'', strings[3], count=subcount[17])

take 140 ms to run

while full benchmark takes 300 ms.