Issue found and solved!
Setup:
4x Couchbase Server 4.0 CE
4x SG 1.2
I restarted all nodes one after another and the issue is gone. Looking at the Android monitor now there is way less network activity now, and because of that there is way less CPU and RAM activity, and the lag is gone. The VMs were running for weeks without issues but I strongly believe that restarting solved the issue. That the issue was that CBL and SG or Couchbase server were communicating a lot because of an unknown issue.
Would it be helpful to send any SG logs or Couchbase server logs so that someone can investigate?