domain_id based replication filters

Provide a mechanism to filter replication events on the slave replication
server based on domain_id.

New master definitions:

(1) DO_DOMAIN_IDS = (N[,N])

• Accepts a comma-separated list of domain_ids, the replication events of
  which will be kept.
• If a CHANGE MASTER command is executed without DO_DOMAIN_IDS, the
  existing list of domain ids (if any) would be preserved.
• DO_DOMAIN_IDS=() can be used to clear the list.
• The list should show in SHOW MASTER STATUS output and master_info file.
• On restart, the list should automatically get populated from master_info.

(2) IGNORE_DOMAIN_IDS = (N[,N])

• Accepts a comma-separated list of domain_ids, the replication events of
  which will be filtered out.
• If a CHANGE MASTER command is executed without IGNORE_DOMAIN_IDS, the
  existing list of domain ids (if any) would be preserved.
• IGNORE_DOMAIN_IDS=() can be used to clear the list.
• The list should show in SHOW MASTER STATUS output and master_info file.
• On restart, the list should automatically get populated from master_info.

Additional notes:

• The list preservation rule is consistent with IGNORE_SERVER_IDS.
• Both the lists (DO_ & IGNORE_) cannot be non-empty at the same time.
• CHANGE MASTER TO command would fail with ER_MASTER_INFO if it tries
  to set both the lists.

Examples:

CHANGE MASTER TO .. IGNORE_DOMAIN_IDS= (1,2,3) .. ;
CHANGE MASTER TO .. DO_DOMAIN_IDS= (99) .. ;
CHANGE MASTER TO .. DO_DOMAIN_IDS= () .. ;

$ tail -2 master.info
do_domain_ids=0
ignore_domain_ids=3 1 2 3