Show Progress of Execution More Precisely And with More Granularity

[imans777]

Hi,
Currently, the progress of execution is calculated by the size of replicas that their movement is already finished.
But suppose we have a large cluster that has replicas of 500GB order. In that case, when moving 5 replicas, it takes more than 8 hours (in our environment) to complete. In this time, anyone checking the progress through cruise control rest API, they'll see that the progress is 0 MB, and when they all complete, it suddenly changes to 2.5TB.
In the meantime, if we want to make sure that the progress is actually running and nothing is troubling, we have to go and check disk usage and see that their size is changing (and of course, see that how much of the replica is moved/left to move), which can be really headache in clusters with hundreds of brokers and disks.
This is also important when we're playing with throttling parameter; we might set a value that the replication becomes so slow that might become infinitely, that way we would realize this fact after maybe several days.
Also it would be pleasant for CC to estimate the remaining time, based on initial parameters and then the progression.
Thanks.