Fixes #1854

Also fix bug where bogus charset names like `iso` were confused with valid
charset names starting with that text, such as `iso-8859-2`.

Ensure that the References header that is included with outbound email
definitely includes the References header from an email received into the
system, if at all possible.

When the user is not identified by the web interface, such as processing
email to tickets, attempt to locate and include the user filter information
when matching the filter to the new ticket data.

If an update to an object such as a ticket results in a corresponding update
to the CDATA table, and one or more of the fields have been deleted from the
ticket details form, then, avoid updating the CDATA table for those fields.

This fixes an issue where the CDATA table is dropped and recreated when
fields are added and removed from the form. The table will only contain
fields which are currently on the form when the table is created. Therefore,
deleted fields will never be on the CDATA table and will result in an SQL
error.

This patch adds an error to the page if unable to configure the archive
folder and POP was selected as the mailbox protocol. Generally, POP servers
do not support anything but the INBOX

Unassign tickets on transfer when the target department has assignment
restriction and the assigned staff is not a member.

Disable claim (quick self-assignment) when above restriction is in effect.

Here's an example email structure:
```
multipart/mixed
 - multipart/report; delivery-status
   - multipart/alternative
     - text/plain
     - text/html
   - message/delivery-status
   - message/rfc822
```

The previous code would only find the body if the email main headers had:
Content-Type: multipart/report; report-type="delivery-status". In such a
case it would scan for a plain/text body.

This patch will scan for the usual body if the scan for the body as usual if
the report scan did not find anything.

Also, output errors to stderr when running API from the command line

It seems that most systems handle the newline character alone without a
problem; however, some systems cannot handle the standard CRLF line ending
at all.

The %ticket_email_info table needs an index on thread_id. This is done in
the `develop-next` branch; however, this patch helps alleviate performance
issues on the main branch until the `develop-next` branch is collapsed for
main line development.

Fixed typo

This also adjusts the logic internally to lookup the priority by the
priority_urgency field, which should be constrained to follow priority
precedence over the priority_id field.

Because CSV has no way of indicating a content character set, writing the
file with a Unicode byte-order mark is useful to declare the content in
UTF-8 encoding.