Move nodes_monitor builders to scrubber and improve keys_lock timing logic

[jeniawhite]

Environment info

• Version: VERSION
• Deployment: AZURE|ESX|AWS|GCLOUD|DEV|VBOX
• Customer: NAME

Actual behavior

• Currently we use the builder from scrubber and from nodes_monitor
• We use the builder in nodes_monitor in order to show the status of the rebuild to the UI (we actually calculate the percentage of rebuild)
• I've created a class that will handle the locking of chunks in the builder in order to fix an issue when we call to lots of builders on multiple nodes rebuild from nodes_monitor and they are working on the same chunks
• Even with this code since the scrubber and the nodes_monitor builder are running from different processes, we still have the same problem of timing between them.
• The current code only fixes the timing issues relevant to the webserver (nodes_monitor) process
• Also the algorithm that handles the locks of the chunk batches is not optimized to the fullest

Expected behavior

• We should optimize the algorithm and move the builders from nodes_monitor to the scrubber
• That way we will fix the timing issues between the scrubber and the nodes_monitor as well

[guymguym]

Landed in 2.0 (08d9022)