faster cython version for bytes

[zed]

def count_doubles(bytes data):
    cdef Py_ssize_t count = 0, i
    cdef char *s = data  # no copy
    for i in range(len(data) - 1):
        count += (s[i] == s[i + 1])
    return count