Approx. 50% less memory footprint and 40% improvement in query time

In the best case (all ascii), this reduces the memory footprint by 60%
and the response time by 15% to 20%. In the worst case (every line has
non-ascii characters), 3 to 4% overhead is observed.

Close #513

This change improves sort ordering for aligned tabular input.
Given the following input:

    apple   juice   100
    apple   pie     200

fzf --nth=2 will now prefer the one with pie. Before this change fzf
compared "juice   " and "pie     ", both of which have the same length.

    > time cat /tmp/list | fzf-0.10.1-darwin_amd64 --ansi -fqwerty > /dev/null

    real    0m4.364s
    user    0m8.231s
    sys     0m0.820s

    > time cat /tmp/list | fzf --ansi -fqwerty > /dev/null

    real    0m4.624s
    user    0m5.755s
    sys     0m0.732s

    > wc -l /tmp/list2
     2594098 /tmp/list2

    > time cat /tmp/list2 | fzf-0.10.1-darwin_amd64 -fqwerty > /dev/null

    real    0m5.418s
    user    0m10.990s
    sys     0m1.302s

    > time cat /tmp/list2 | fzf-head -fqwerty > /dev/null

    real    0m4.862s
    user    0m6.619s
    sys     0m0.982s

I profiled fzf and it turned out that it was spending significant amount
of time repeatedly converting character arrays into Unicode codepoints.
This commit greatly improves search performance after the initial scan
by memoizing the converted results.

This commit also addresses the problem of unbounded memory usage of fzf.
fzf is a short-lived process that usually processes small input, so it
was implemented to cache the intermediate results very aggressively with
no notion of cache expiration/eviction. I still think a proper
implementation of caching scheme is definitely an overkill. Instead this
commit introduces limits to the maximum size (or minimum selectivity) of
the intermediate results that can be cached.

Close #193