This won't impact Synchronet as it has a separate signal handling
thread, but we still need to behave properly for processes that
don't.  I'm also saying that ENOMEM does not indicate a disconnection,
though it may be better to pretend it was disconnected...

The goal of this commit is that: only modules that either are part-of sbbs.dll/libsbbs.so or need to link with/use that library, will #include "sbbs.h" and thus be dependent on its subsequent includes (e.g. cryptlib.h, jsapi.h). This should mean extdeps.mk can be trimmed way down.

I also removed CVS keyword/comments and trimmed up the boilerplate copyright notice in modified and added source/header files in this commit.

There is no functional change in behavior in this comment.

This means the JS global word_wrap() method has a new optional Boolean arg
as well.

Most of the copyright years in the source code were misleading (the date of
most recent publish was actually later) and all were unnecessary. I've been
removing copyright years piecemeal, for a long time, but I decided it was time
to just perform a bulk search and (mostly) replace. In some cases, I left
old copyright years on files that either are not used (and soon to be removed)
or obsolete and unlikely to ever be touched again (e.g. Win9x FOSSIL VXD). Some
of the runtime binaries still contain copyright years and those were updated to
2018.

"Change the last argument of wordwrap() to a flags argument and support
making a bare LF not be ignored."
If soft line breaks (bare LFs) are stored in the message base, they're
expected to wrapped for 80-column terminal only.

Hopefully I did this right (Deuce?).

Separated code from crc32.h into crc32.c (to allow combination of mailsrvr and sbbs into single module).