It is almost 2024 after all. :-)

Every server should have *some* limit to protect against DOS attacks.
Every connected client consumes a socket, a thread, some memory, none of
which are infinite resources.

Put th row inside a thead as is proper.
Put td rows inside a tbody as is proper.
Close td elements even if not strictly necessary. Good form.
Use a subjectively nicer font.
Use a subjectively more readable date format.
Put some space and border between table elements.
Stripe rows for better visual indicator of separate data.
Make the header row sticky so we always know which column's what.
Add dark mode colors if browser prefers-color-scheme dark.

... unless the full path was specified.

Also, expand the lock in websrvr to the correct scope.

If this solves the issue, it means that all certs for a specific
scfg will share the same compliance level.  For the mail server,
this level is OBLIVIOUS, so if it smells like a cert, it's considered
valid.

If this doesn't solve the issue, it needs to be backed out.

May help nail down issue with keys.

We'll hold a reader lock under the session is established, which
should prevent blocking other threads unless something is beating
on get_ssl_cert() (which would be a different bug).

This still needs to be figured out, but at least this should fix
the immediate issue.

Asked for a way to expand list of recipients for a specified user

Remove some redundant redundancies, redundantly

it's only called from load_cfg() which has already set it to -1.

We also don't need to check prepped twice to make sure it's extra true.

The design is to #include eventwrap.h for cross-platform source

Ctrl-A in the Failed Logins dialog now selects-all rather than copy-all (just
use Ctrl-A, then Ctrl-C to copy all).

warning: result of comparison of constant 100000 with expression of type 'uint16_t'
(aka 'unsigned short') is always true

./scfg.h:80:9: warning: 'SUB_HDRMOD' macro redefined [-Wmacro-redefined]

This is always going to call the terminal server's lprintf function (when used
with libsbbs.so/sbbs.dll) which is probably not what was intended.

Resolve reported clang warning:

inkey.cpp:592:20: warning: implicit conversion from 'long' to 'int' changes value from 2147483648 to -2147483648 [-Wconstant-conversion]
         if(!term_supports(MOUSE))
             ~~~~~~~~~~~~~ ^~~~~
 ./sbbsdefs.h:574:19: note: expanded from macro 'MOUSE'
 #define MOUSE           (1L<<31)                /* Mouse supported terminal                             */

This is hacky cruft here. We shouldn't have the declaration in multiple
places. It probably shouldn't be in sbbs.h.