We initially implemented the algorithm where the number of sections
is passed in as a parameter. However, it's not the number of sections,
it's the length of each section.

This was an easy fix, as the algorithm is pretty symmetric.

The right option to mimic the given example is -s 2, not -s 4.
(But curiously, this leads to the same answer).

In particular, a case where the target string is different from
all the strings:

    1110
    1101
    1011

If we flip all the 0 bits to 1, all the strings will be equal (1111),
but 1111 is not one of the given strings.

Since they need different command line options