strrchr() here to find terminating chars (quotes, parens, brackets).
e.g. '"Joe Shmoe" <joe@shmoe.com>, "Fred Smith" <fred@smith.org>'
would be parsed as 'Joe Shmoe" <joe@shmoe.com>, "Fred Smith'.

'data_length', but not always (e.g. may include text from some header
fields, excludes any non-text data fields).
Mark 'data_length' msg header property as read-only.

back in 1993, but were never needed/used.

header fields. Eliminate smb_getcontenttype() - now unnecessary.

the string being passed around.

or NULL if not relevant (not a MIME-encoded message).
Added smb_countattachments(): returns the number of file attachments in a
MIME-encoded message.
Added SMBMSGTXT_NO_TAILS definition to be used in place of '0' for
common smb_getmsgtxt() mode argument value.
The amount of white-space after "Content-Type:" and the content-type value
is now more flexible (0 or +1 spaces/tabs are fine).

sendfile() no longer logs "attachment" when a file description is provided.

download_file_attachments() and expose via new JS "bbs" method. Currently
this only reall works for the mail base.

since the smb_t already contains a subnum and we were passing a pointer to the
sbbs_t::smb to lower-level called-functions anyway.
Let's not mix-up class-scope with function arguments.

Fix the missing carriage-returns in the shell (bash) prompts by always building with the stderr logging
support (in *nix builds) - but run-time disable as desired via EX_NOLOG option. I've been wanting to get
rid of that "XTERN_*" macro for a long time anyway (it's "XTRN" damn-it!). :-)

Added EX_NOLOG to *nix shell mode - don't log the shell output.

- Renamed EX_INR and EX_OUTR to EX_STDIN and EX_STDOUT
  (with duplicate variables for backwards compatibility)
- Removed long-obsolete EX_OUTL, EX_SWAP
- Added (new) EX_NOLOG

intercepted output to stderr won't be logged.

Since QUOTEWRAP *is* used, just define that bit flag specifically.

list (smb_list) for thread-safe SMB stack storage.

and frankly could cause all kinds of havok with multiple thread (e.g. nodes)
pushing and popping SMB's concurrently. eek.

More use of get_index() if available, to help speed things up a bit.  Also, made a further optimization: when checking a sub-board if it has polls, check in reverse rather than forward.  Since polls & voting is a relatively recent feature in Synchronet, hopefully it should finish faster going in reverse.

SlyVote v1.02: Now uses the new get_index() messagebase function, if available, for getting the message index objects when determining if a sub-board has polls (in the subBoardHasPolls() function).  get_index() is faster than iterating through all messages and calling get_msg_index() for each message.

of the message header so that functions that need it (e.g. bbs.show_msg_header)
don't use/show stale data. This change requires the updated smb_copymsgmem()
which clear/sets message convenience pointers upon copying hfields from one
message header to another.

Did not clear/set convenience pointers (e.g. msg.subj), so they would point
to the source message hfield_dat buffers which may be freed or garbage.
Apparently nothing that used smb_copymsgmem() used the convenience pointers
(before now), so it hasn't been a problem before now. <shrug>
It became a problem becaue I'm going to use this function in js_put_msg_header
to fix a problem where the underlying smbmsg_t representation is not current
even after MsgBase.put_msg_header() is used.

(objects, in the same format as returned by the get_mgs_index() method)
much faster than iterating through a loop, calling the get_msg_index() method
for each message. If you want to load messages (e.g. headers, text), filtering
by criteria found in the message index (attributes, to/from user, subject CRC)
loading a list of indexes and filtering before calling get_msg_header() for the
selected messages is much faster than previously available MsgBase object
methods (e.g. get_all_msg_headers()). If you don't need to filter the loaded
messages (you really want *all* message headers), then get_index() is of no
benefit to you, the script-writer. This is most useful for the "mail" msgbase
where selective loading of messages is more common.

Ctrl-A (ASCII 1) char. This is wrong. A literal Ctrl-A sequence is ^AA.
I'm only addressing one of those instances in this commit since it *also*
appears to be an off-by-one bug where it's treating *all* Ctrl-A sequences as
though they were ^AA - so just #ifdef'ing out until Deuce takes a look at this.

blank line) sequence.

(use the sysop 'D'elete command to quickly delete a range of messages).

js_put_msg_header (MsgBase.put_msg_header) - we can't free the memory
associated with 'gotten' msg header. Just use the header offset from the
header retrieved via MsgBase.get_msg_header() - for fast lookup.
Also fixed an innocuous extra increment of 'n' in argv/argc loop.

Updated help text and option name, a bit.

Linux system limitations and user/port configurations of sbbs.
Add some parens around bit-wise "and" (&) checks too.