Our servers were blowing up with OOM.

I was doing regression tests to know what changes were responsible of the leaking. WE found them and fix them

I worked with horacio to polish the new feed API behavior. Now we have mentions insithe the feed, with references to source and target resources and everything paginated. Account for moved documents and deleted documents as well.
Also I was improving unitary tests to spot issues in case we make another mistake in the code.

The plan for the next week is to finally start working on the pushing algorithm we greenlighted last week and keep supporting horacio in the Feed API