Changed is_open() to work as before.
Added back inited argument to LOG


mysql-test/r/rpl_flush_log_loop.result:
  Fixed results (probably bug in previous rpatch)
sql/handler.cc:
  Changed is_open() to work as before
sql/item_func.cc:
  Changed is_open() to work as before
sql/log.cc:
  Part revert of previous patch.
  The reason for adding back 'inited' is that is that we can't be 100 % sure that init_pthread_objects() is called before mysqld dies (for example on windows)
  I removed mutex lock handling in is_open() as the new code didn't have ANY affect except beeing slower.
  Added back checking of is_open() to some functions as we don't want to do a mutex lock when we are not using logging.
  Indentation/comment fixes
sql/log_event.cc:
  Changed is_open() to work as before
sql/repl_failsafe.cc:
  Changed is_open() to work as before
sql/sql_base.cc:
  Changed is_open() to work as before
sql/sql_class.h:
  Changed is_open() to work as before. Added back 'inited' variable
sql/sql_db.cc:
  Changed is_open() to work as before
sql/sql_delete.cc:
  Changed is_open() to work as before
sql/sql_insert.cc:
  Changed is_open() to work as before
sql/sql_load.cc:
  Changed is_open() to work as before
sql/sql_parse.cc:
  Changed is_open() to work as before
sql/sql_rename.cc:
  Changed is_open() to work as before
sql/sql_repl.cc:
  Changed is_open() to work as before
sql/sql_table.cc:
  Changed is_open() to work as before
sql/sql_update.cc:
  Changed is_open() to work as before