Since lispflowmapping works well with jdk8, I see no reason to have
different JDKs for Beryllium and Boron+, which is another variable
that's different between the tests. While it should not siginificantly
affect performance, I would feel better if we ran all tests with the
same JDK.
Change-Id: I5f4b239064f6c3bff7f00640c3eab01bf6051b85
Signed-off-by: Lorand Jakab <lojakab@cisco.com>