Index creation should be started after reindex finish

Right now index creation process is started (optionally, basing on consumbed blocks number and psql-index-threshold parameter) when hived node enters live sync. This is too late and disallows effective regression testing of applications, where we're performing a replay up to specified number of blocks. Above should not break massive P2P sync. psql-index-threshold checks could be also specific to estimated block distance to real blockchain head block (basing on current real time and time of currently consumed head block by given node).