router-perf add test_routes after mb test for a while to check service health

[qiliRedHat]

1. After mb test, add another test_routes, to check if all the routes are still be able to work. If the last test_routes fails, the test should fail and not delete the objects, so we can check what happens to the cluster.
2. Add a new env variable to wait for specific time before doing the last test_routes. This is aimed to cover the test scenario that all routes are still working after the cluster long run for a while, for example 1 hour.

@rsevilla87 This is a test requirement from OCP network QE team. Please let me know what you think. I can do the code change after we get to an agreement.
CC: @mffiedler