Improve wc

[gcanat]

As suggested in #137, I took inspiration from wc2 to improve wc_file_bytes.

In my benchmarks (tested with 1G utf8.txt file generated with wctool), it is a bit faster than wc2, and on par with wz, except when passing multiple files as argument because wz will parallelize computation, but I guess we don't want to pull rayon as a dependency here :)

[jgarzik]

Looks good at a quick glance.

The table initialization can be improved further with a static table, yes? e.g. https://www.reddit.com/r/rust/comments/1dokkh2/recipe_for_compiletime_sparse_array_initialization/

[gcanat]

OK I made the change.
For what it's worth here's the result on 1G utf8.txt

tool	branch	time
posixutils-rs	main	2.66s
posixutils-rs	wc2	1.50s
wc2.c	main	2.10s
wz	main	1.57s
wc	v8.32	15.33

[jgarzik]

Thanks, merged.

Interested in tackling wc_file_chars()? :)

[jgarzik]

Post-merge comment: was_space need not appear in CountInfo at all, as far as I can tell.

[gcanat]

Oh yeah, I can have a look at wc_file_chars() in the coming days. And will try to handle was_space differently.

[jgarzik]

Oh yeah, I can have a look at wc_file_chars() in the coming days. And will try to handle was_space differently.

Thanks.

re was_space, it appears just a local, since it is not accumulated, and should not count across different files.