We've started exceeding the 6.5h timeout for centos regresssions. Let's bump it up across the board until we can get chunked regressions going.
Done: https://review.gluster.org/#/c/18705/