SIGWINCH support

[mptre]

No description provided.

[mptre]

Any objections on using C99? /cc @calleerlandsson @mike-burns

[ghost]

This will break my beloved gcc 2.95 on ancient machines here ... but anyway: time moves on ... and all this just for including some stuff from OpenBSD-folks, that no one uses ... ;-)

[mptre]

@DBOTW can you confirm that compiling this PR with GCC 2.95 still works?

[ghost]

After some more "tweaking" (replace -Wextra with -Wno-format) and defining LLONG_MIN and LLONG_MAX in compat-strtonum.c (I knew this would cause trouble ;-) it works!

[mptre]

@DBOTW thanks for confirming.

[mptre]

The remaining pieces to get SIGWINCH working is now merged into this
PR.

Does anyone have access to a machine where SIGWINCH is not defined?
I'm more in favor of skipping the autoconf check if we can't test it
properly. Even the tmux portable unconditionally assumes its
present as far as I see it.

[ghost]

Hard to find such a machine: even HP-UX 8.x (and earlier) have SIGWINCH (called SIGWINDOW) ...
I fired up my HP9000 Model 715/50 with HP-UX 10.20 and compiled pick with SIGWINCH support and it worked ... most of the time ... sometimes pressing ^L was necessary to clean up the mess ... :-)

Seems this was caused by a missing

new_attributes.c_cc[VMIN] = 1;

in tty_init(void) ... now everything runs perfect (even resizing) under HP-UX 10.20:

[mptre]

Thanks for checking! I then advocate on omitting the autoconf check for
now and get this released. If we receive any reports on build failures
we could revisit this matter when we know it's an actual issue.

[calleluks]

This looks great @mptre and @DBOTW. Thanks for all your great work! Feel free to rebase and merge, @mptre.

[mptre]

Thanks everyone, merged!