Log a warning message when either the "local" or "in-transit" flag has been
stripped from ("Sanitized") imported packed-messages.

Keyop reported a chksmb warning on a FidoNet message that contained the
'in-transit' attribute:
<Keyop> Message flagged as 'In Transit'     (?): 1
<Keyop> fidonet_fidosoft Message Base has Errors!

The message in question appears just to be a corrupted echomail msg coming
from HPT:
Sender           '56:43 '
To               Tommi Koivula
Subject          test
X-FTN-AREA       FIDOSOFT.HUSKY
X-FTN-TID        hpt/os2-wc 1.9.0-cur 2021-03-14

SBBSecho didn't use to convert/import the "in-transit" flag on imported
netmail or echomail messages, ... until commit 446ab0ef, which
was needed to support the TickFix robot. So now, let's just sanitize the
packed-message attributes a little sooner in the import process and log a
warning when such sanitization has taken place.

The notification of offline users was removed from the mail server 13 years
ago (commit 729eb16b) without a lot of fanfare or explanation. Add an
option for notification of offline users (in addition to online users) for
those sysops that want that behavior.

Use the userdat is_user_online() function in place of the getnodedat() loop,
which also wasn't considering nodes in "logon" status as having a user online.

Contrary to https://mosquitto.org/api/files/mosquitto-h.html#mosquitto_loop_stop, passing force=true is not enough for the mosquitto network thread to end, mosquitto_disconnect() must be called.

JSDOC fix-ups.

Mostly just bitfield->bit-flags.

A little more visiblity into the MQTT library (version, function return values)

Recently started truncating trailing whitespace from sender/recipient names
before calculating the CRC value (similar to what we do for message subjects
as of a year ago). To suppress errors when evaluating messages (with trailing
whitespace in sender or recipient namems) created or imported before this
change, use this new -N option.

Make it obvious when a message header value string has trailing white-space.
Make this function (and binstr()) thread-safe by eliminating the local static
variable.

This should prevent programs using rgbmap (ie: SyncTERM) from
having an exectuable stack on Linux.

And don't offset long(er) windows from edges to allow more options to be
on screen at once.

Tested with 50 column mode.

Maximize screen realestate for options/settings display:
- Don't place a margin around windows when in a narrow screen mode.
- Don't display shadows on windows when in a narrow screen mode.

This commit also fixes a bug in uinput() when left was non-zero and the total
width (left + width) would be too wide for the screen. This bug was apparent in
SCFG->Message Areas in 50 column mode (no Message Group window would be
displayed when a message group was selected).

Some BBS packages tack on trailing whitespace on names, e.g.
https://github.com/wwivbbs/wwiv/issues/1607

Fixes Windows build.

Right now, only supports publish, subscribe to come next.

Only included with the mosquitto lib support is enabled in the build.

Only on Linux, Documented as VSWTCH (also added), as a SysV-only
thing but actually defined a VSWTC, must be disabled (ie: set to
_POSIX_DISABLE) to not show up in stty as being modified.

*sigh*

I'm not convinced Linux developers want termios to work.

Be sure to set them first though, since they may overlay with
other ICANON entries.

The wrong characters in the wrong order.  I have no idea why they
even have the thing defined, but we'll just stop using it since
it seems BSD systems are the only ones where it's useful...
instead just set V* to C* except ERASE and ERASE2 which we set both
to CTRL-H

The the date/time display would overwrite parts of the application
title in narrower screen modes, so let's not do that.

No functional change, just a little clean-up.

ftouch(), semfile_signal(), and JS file_utime() would fail to update a file's access/mod times with errno=EPERM if not run as the file's owner. From "man utime":

EPERM times is not NULL, the caller's effective UID does not match the owner of the file, and the caller is not privileged (Linux: does not have the CAP_FOWNER capability).

So use a NULL times parameter value when updating to a file's time stamp(s) to the current time.

- getscaling() incorrectly used an int intermediate variable
  This would force scaling to an integer value when read.
- Only parse ConfigureNotify events describing a change to the SyncTERM window
  It appears that XWayland on ChromeOS was occasionally sending a
  ConfigureNotify event describing some other window to SyncTERM,
  and that window had a 1x1 size, which would cause SyncTERM to
  assert minimum size again (usually 640x480), which would set
  scaling to 1.0.

One of the benefits of the new .ini format config files. :-)

This increase introduced a few potential buffer overrun warnings (all writes
to smb_t.file ?)

... using snprintf() instead of sprintf()

Since we are now free()ing the opt[] items in bail(), I've seen heap
corruption core dumps after navigating the SCFG menus with some long
configuration items (e.g. timed events with max-length command-lines).
This heap corruption has been present for a long, long time, but not
detected since we weren't free()ing these heap-allocated items.
Increasing MAX_OPLN would have been a mitigation for this issue, but
using snprintf(opt[n],MAX_OPLN,...) is the better fix.

Copy/pasted (with little modification) from
https://synchro.net/docs/file_section.html#AdvancedOptions -> "Short Name:"

Some short names have specially predefined meaning and it's important to
include that detail here.

Unlike strip_ctrl(), does not remove any other control characters from string.

1) In savescreen(), if the vmode is not found, explicitly set
   pixels to NULL rather than leaving it uninitialized.
2) Ensure that currmode is never left as _ORIGMODE (which is not
   a defined mode, so can result in vmode not being found).