This is that rare case where a Coverity fix makes the code more
obvious and shorter.

This ones does require SSIZE_MAX... so let's see what the pipes say...

These are either correct and trivial to fix, or false positives,
but fixing is easy enough that it's not a problem.

While it's only been around for about 18 years, I'll assume that
anyone updating the exec directory has updated Synchronet as well.

(user_t.misc, User.settings), when enabling auto-terminal detection. This is
similar to what we do in logon.cpp already:

	useron.misc |= (AUTOTERM | autoterm);

We used to have this logic in useredit.cpp, but was removed in commit
8a71ab16, probably due to editing offline users (e.g. in ;uedit sysop
comand). So just do the copy of autoterm flags when editing the *current*
(online) user. This likely addresses the original concern in that commit.

And it fixes issue #853, thanks to Deuce for root-causing

... since that puts us back into VGA tall-font mode

Each app can have its own config file as well (e.g. ctrl/scfg.ini)

We're just converting/upgrading this file (to time.ini), so don't log an error
(reading) if/when the file doesn't exist.

Fixes Error: Unresolved external 'GetNetworkParams' referenced

Some things were using the global i and not a local i... this was
not a problem.

Added alerts when lightbar.draw() aborts due to no ANSI or no colour.

Fixes JPEG XL again.

... or at least, try really hard not to

This is kinda valid... I am reading a 32-bit integer from a socket,
then trying to loop that many times.  What Coverity didn't notice
though is that the loop will exit way before due to the buffer
being exhausted.

By ensuring that it's less than UINT32_MAX/4, I'm still not really
enforcing anything since with the 32-bit value for extcnt, and a
32-bit value for every string, there could only be strings before
it failed anyway, even with a 4GB buffer (which you couldn't do).

This gets the assignment to the rip.* members out of the lock so
Coverity doesn't assume they need to be protected by the lock.

It should compile to the same thing.

They're not "leaking" if I want them to liver forever. :D
This does prevent an extra needless load if User.dll though...
but sdlfuncs and win32gdi both still load the same two DLLs.

since WordPad is no longer included in Windows

Check and log the length of the packet now (since fexist() continues to report
that the file exists when fexistcase() said it did not, even when
experimenting with adding a call to access(filanem, F_OK).

This likely affects no system but VERT.

Cleans-up (deletes) expired filters in text/ip.can and ctrl/spamblock.cfg

Use an atomic that's checked in the while loop in the io threads,
for SSH, we needed to add a new one since ssh_active indicates the
cryptlib session is active, so use ssh_complete to indicate that
a CRYPT_ERROR_COMPLETE (or other final message) occured.

This should fix issue 174.

ircd.js kept crashing on startup on vert with these errors, right after it
logged:
Accepting new connection on port 6667
0832 Accepted new connection: 71.95.196.36 port 55646
[UNREG] Resolving IP: 71.95.196.36

No functional change.

Many sysops probably don't realize these keys do special stuff.

so we can use it in nopen.c without requiring sbbsdefs.h ... smbdefs.h

This should fix the SVDM build.