Add NEWS from releases to master branch too

@@
expression E;
@@

- if (hash_table_is_created(E)) {
-       hash_table_destroy(&E);
- }
+ hash_table_destroy(&E);

lib-sieve: redirect action: Assert that dupeid is not NULL when act_redirect_get_duplicate_id() is successful.

Addresses scan-build report.

lib-sieve: redirect action: Fix lack of NULL checking in new X-Sieve-Redirected-From header comparisons.

Problem found by scan-build.

In this context, it can never be NULL and the check confuses Coverity.

In this context, it can never be NULL and the check confuses Coverity.

Also check the X-Sieve-Redirected-From header for our own e-mail addresses. This
header is added by the redirect action itself and in a mail loop it would see
that same header with that same content. This is less reliable than the other
mail loop detection (sender may set such a header), so, unlike the existing loop
detection based on the duplicate db, the implicit keep is not canceled when the
new loop detection is triggered.

Serves as an abbreviation.

It was errorneously using the (potentially) modified mail struct for error
reporting.

This prevents users from messing with redirect loop detection.

plugins: sieve-extprograms: Use sieve_ prefix consistently for sieve elements visible as exported symbols.

This fixes a symbol clash with the imap-filter-sieve plugin. Both modules had
the symbol "cmd_filter".

plugins: imap-filter-sieve: Properly discard the originally stored message when a modified version is stored by Sieve.

The Sieve interpreter can return a flag that indicates whether the original
message should be kept, but that was not actually being used.

plugins: imapsieve: Properly discard the originally stored message when a modified version is stored by Sieve.

The Sieve interpreter can return a flag that indicates whether the original
message should be kept, but that was not actually being used.

Instead use the detected ones.

Otherwise stats process startup can get into a loop.

Coverity complains about last_script possibly being empty in
imap_sieve_filter_run_scripts(), which is actually not possible, since the
function would not be called if there is no script to execute. Added assertion
to make that clear.

The debug message pertains to skipping secondary scripts that failed to
compile. This debug message will not actually be triggered in the current
implementation, but this will become problematic in the future.

Problem found by Coverity.

Compile errors occurring for Sieve scripts uploaded in multiple TCP frames were
not handled correctly.

The IMAP FILTER=SIEVE capability does not execute more than a single script yet,
but once it does, it should not assert fail on secondary scripts that failed to
compile.

lib-sieve: util: rfc2822: Prevent writing header lines with trailing whitespace in rfc2822_header_append().

Panic was: "Buffer write out of range"

With some rather weird (sender-provided!) input, the header folding algorithm
got confused, causing a pointer to the start of the current line to exceed the
parsing pointer. This caused str_append_data() to be called with a negative
size. Added an assertion to make any future similar problems more obvious.

After finishing reading the Sieve script, the command parsing sometimes didn't
continue with the search arguments. This is a time-critical bug that likely
only occurs when the Sieve script is sent in the next TCP frame.

Occurs when several mails from different backend mailboxes are involved in the
COPY event. Fixed by using mail_save_context->>copy_src_mail in mailbox_copy()
instead of the mail argument. The latter can point to the backend mailbox for
virtual mailboxes, which makes no sense to the imapsieve plugin as it expects
only one source mailbox.

Panic was:

Panic: file imap-sieve-storage.c: line 337 (imap_sieve_add_mailbox_copy_event): assertion failed: (ismt->src_box == NULL || ismt->src_box == src_mail->box)

Change was:

lib-mail: message_address_parse() - Change fill_missing parameter to flags