misc replication bugfixes including some needed modifications in IO_CACHE

likely() and unlikely() branch prediction compiler hint macros
clean-up of comments


include/my_global.h:
  added likely() and unlikely() macros to help some compilers optimize
  the code for architecture-specific branch prediction policies
include/my_sys.h:
  coverted my_b_append_tell() from macro to a function as it needed to be more
  complex to avoid a potential race condition
mysql-test/mysql-test-run.sh:
  hostname-independent relay log name to have consistent SHOW SLAVE STATUS
  output
mysql-test/r/rpl000014.result:
  result update
mysql-test/r/rpl000015.result:
  result update
mysql-test/r/rpl000016.result:
  result update
mysql-test/r/rpl_log.result:
  result update
mysql-test/t/rpl000017-slave.sh:
  proper cleanup of old logs
mysys/mf_iocache.c:
  cosmetic changes + more debugging asserts
mysys/mf_iocache2.c:
  my_b_append_tell()
  cleanup of comments
sql/log.cc:
  fix potential bug - do not rotate log in the middle of event
sql/slave.cc:
  do not write stop events when the server does not actually stop but just 
  rotates the log
  fixed race between queue_event() and show_slave_status()
  clean-up of comments
sql/slave.h:
  added ignore_stop_event flag to SLAVE_LOG_INFO

mysql-test/mysql-test-run.sh

@@ -832,6 +832,7 @@ start_slave()
   slave_args="--no-defaults $master_info \
   	    --exit-info=256 \
           --log-bin=$MYSQL_TEST_DIR/var/log/$slave_ident-bin \
+          --relay-log=$MYSQL_TEST_DIR/var/log/$slave_ident-relay-bin \
           --log-slave-updates \
           --log=$slave_log \
           --basedir=$MY_BASEDIR \