backport the new fix from 10.0

1. move cflags/libs cleanup from mysql_config.sh (runtime)
   to for_client.cmake (build time). Include/library paths are
   still calculated at runtime (they depend on the location of mysql_config)

2. Use cleaned-up cflags/libs to generate mariadb.pc

3. remove obsolete @expansions@ from mysql_config (for variables that are
   never set in cmake files)

add bsdutils dependency, because mariadb-server-5.5 deb package uses logger

apply debian's patch to workaround kfreebsd bug

and apply whatever was reasonable

typo (thanks, Brad Smith!)

stat structure (from <sys/stat.h>) is conditionally defined
to have different layout and size depending on the defined macros.
The correct macro is defined in my_config.h, which means it MUST be
included first (or, at least before <features.h> - so, practically,
before including any system headers).

            SET @@global.log_output

Deadlock chain:
rdlock(LOCK_logger) -> lock(LOCK_open)     SELECT 1
lock(LOCK_open)     -> lock(LOCK_status)   DROP TABLE t1
lock(LOCK_status)   -> lock(LOCK_g_s_v)    SHOW STATUS
lock(LOCK_g_s_)     -> wrlock(LOCK_logger) SET @@global.log_output=DEFAULT

Fixed by removing relationship between LOCK_status and
LOCK_global_system_variables during SHOW STATUS: we don't really need
LOCK_global_system_variables when accessing status vars.

            SET slow_query_log and failed connection attempt

A very subtle though valid deadlock. Deadlock chain:
wrlock(LOCK_grant)    -> lock(acl_cache->lock) GRANT/REVOKE CREATE/DROP USER
lock(LOCK_open)       -> rdlock(LOCK_grant)    SELECT * FROM I_S.COLUMNS
wrlock(LOCK_logger)   -> lock(LOCK_open)       SET @@global.slow_query_log='ON'
lock(acl_cache->lock) -> rdlock(LOCK_logger)   Failed connection

Fixed by removing relationship between acl_cache->lock and LOCK_logger
during failed connection attempt.

mysql-test/r/func_group.result:
  Test case
mysql-test/t/func_group.test:
  Test case
sql/item_sum.cc:
  Restore ORDER for prepared statements

and other fixes for gcc-4.9.1 on sid

don't install them at all

don't send an OK packet if the SHOW PROCESSLIST was killed

* remove bundled jemalloc, use the system one
* force jemalloc in release builds on linux

bits were inconsistent though the handle was opened without those bits
set. This caused dbremove to always fail.