I'm doing builds for my jail (which doesn't have libmosquitto) on
my system that does have it now, so I need to disable it manually.

Mostly (all?) about unchecked return values. Unexpected failures to read or
write some files could've definitely led to some weird bugs.

Don't null-deref h_addr_list.
Return in_addr_t instead of u_long.

warning: ‘maxval’ may be used uninitialized
warning: ‘inch’ may be used uninitialized

Don't deref a null pointer and return an in_addr_t instead of ulong.

Was wrong and therefore causing flickering.

Was not using struts, and was scaling to window size, not aspect
ratio size.

Also, reset struts when not in fullscreen mode.

GDI and X11 still can't be *started* in fullscreen mode though.

Win32GDI currently requires two rects to be available.

- Ensure last is cleared even if we manage external scaling without Xrender
- Free the bitmap rectangle in the same situation
- Make a copy of x_internal_scaling so it's used consistently
  (not mutex protected, and accessed from multiple threads)

If there's already an outstanding rectangle when the blinker thread
wants to send another one, skip it.

This keeps the interface responsive at the cost of some skipped
frames.

ChromeOS (as of Jun 2023) effectivly makes PRIMARY useless.  Just
do the wrong thing and use CLIPBOARD instead.  This makes it much
more likely that SyncTERM selections will clobber CLIPBOARD
unintentionally, but we have no choice... DigitalMan doesn't use
X11 mode solely because ChromeOS sucks at X11.

While we're hacking on that, make a first attempt at a generic
Atom access framework.

The only actual visible change that should occur here is that the
window will not resize larger than will fit on the current desktop
between various WM widgets (panels and menus).  I've been considering
this change for a while, but now that fullscreen support is present,
it doesn't really take anything away.

Just use _NET_WM_STATE protocol for fullscreen... messing around
with the methods is pretty painful.

Because we're not messing with other stuff, we can likely remove
all the code I added to prepare for this. :D

Testing this really highlighted other broken bits, so a bunch of
that has been fixed as well... including the bug that Ragnorok
hadn't reported yet as of last time I looked (corrupted screen when
maximixed).

Or even weirder, u_long?

And dereffing a ulong/u_long pointer where you expect to find an IPv4 address?
Yet even weirder still.

Fix that spit: It appears in_addr_t is defined on all platforms (?), so use
that type instead.

No known sightings of these sites actually being the location of a segfault,
but as we learned from the segfaults in rblchk(), the first entry in the
h_addr_list can be NULL in some cases.

I'm not sure why this one only started popping up now, but h_addr_list is a
NULL-terminated list and it makes perfect sense that the first entry could
be the NULL-terminator.

gethostbyname is obsolete/deprecated and we should address that in a separate
commit.

A bunch of possible (but often, not really) use of undefined values.
Some ignored return values (e.g. of chsize/ftruncate, read, write, fgets).

Other than some added diagnostics upon some of these unexpected syscall
failures, there should be no change in behavior from this commit.

This is the last feature that SDL mode provides that X11 mode is
lacking.  I rarely use it myself, but it should be there.

Apparently the mosquitto.h won't always be installed in /usr/include on
all systems (e.g. Deuce's FreeBSD gitlab-runner system)

Previously, if there was more than one work area, this would be
an infinite loop, re-reading the first workarea repeatedly.

Reported by Ragnarok (thanks!)

Fix infinite loop in bitmap_drv_init_mode() if scaling results in
larger than max size.

This uncovered an issue in resize_window() which would cause the
screen to be corrupted due to vstat not agreeing to the actual window.

This was ending up using an uninitialized value