Make no of open containers in SCM per Datanode a function of no of data disks

Currently, the no of open containers is based on a config which is per pipeline on a datanode. Ideally, on a given datanode , there can be as many containers open as no of data disks. The general idea is to make to the no of open containers a function of no of data disks on the datanode and then divide it among the no of pipelines a datanode has.