• If the DML activities are not on the historical data (hot dataset and cold dataset), then shall we create a empty partitioned table and start populating the historical datasets into this new partitioned table using LOAD DATA INFILE method. {Preferably we can do this activity in Slave DB. Even we can do this activity in master if we can exclude this new db from replication (created for partition purposes)}. So this will go in-parallel until we reach the hot dataset, that we can do one shot during any planned downtime. I believe this won't take much time as compared to your initial estimation of 24+ hours or so because we are doing the phase by phase data migration. 

"Normally, a long-running ALTER TABLE or other query will stall all following transactions, causing the slave to become behind the master as least as long time as it takes to run the long-running query. By using out-of-order parallel replication by setting the replication domain id, this can be avoided. The DBA/application must ensure that no conflicting transactions will be replicated while the ALTER TABLE runs."

How safe its to alter table on slave and promote slave as master?