Replication filter producing unecessary table locking on the slave

Master have multiple databases with the same non transactional table (blackhole) inside every schema .

We randomly insert records into all databases tables.

Start a replication and filtering on a single database (replicate-do-db=schema1) . We observe sql thread locking all the tables of all the databases . The SQL thread is in locking state on the COMMIT statement . Despite this the filter works for DML and no records are inserted.

#130918 13:45:23 server id 3164  end_log_pos 925        GTID 0-3164-275862697

/*!100001 SET @@session.gtid_seq_no=275862697*//*!*/;

BEGIN

/*!*/;

# at 925

#130918 13:45:23 server id 3164  end_log_pos 1594       Query   thread_id=15    exec_time=0     error_code=0

use `ccmstats_shard04`/*!*/;

SET TIMESTAMP=1379504723/*!*/;

insert high_priority ignore into `ccmstats_shard04`.`ccmreferers`(`ip`,`date`,`firstseenon`,`keyword`,`domaine`,`referer`,`keyword_crc64`)values(1770066430,'2013-09-18 13:45:23','/download/telecharger-34056887','','www.commentcamarche.net','http://www.google.com/url?sa=t&rct=j&q=',5255623658049533571)

/*!*/;

# at 1594

In multi source we have 24 shards so all domain concurrently replay same lock and unlock tables decreasing the all replication speed .

Can we do something to remove the lock introduce by the master when inserting into non transactional table . or can we detect that the lock is part of a transaction that is fully filtered