Changes to support client-server and standalone stress tool runs

[bssrikanth]

This commit has few changes[1,2] required to support running
stress workloads in server-client mode as well as standalone mode
in parallel. Change [3] is a minor fix for handling stress unload.

1. Rename param which is used to specify custom server-client pairs.
2. Now users can run stress tools which are client-server based as well
   as standalone stress tools together. So place existing check under condition
   when user chooses only 'client_server' type of stress load.
3. In case of stopping stress app, in some cases have seen delays to stop
   stress app, giving couple of seconds sleep after stop_cmd fixes the issue

new param which can be used in testcases to configure stress run mode
is documented in base.cfg.

Signed-off-by: Srikanth Aithal [email protected]

[bssrikanth]

Test case would be posted to tp-libvirt.
Below are results:

avocado run multivm_loadstress --vt-type libvirt --vt-extra-params vms="vm1 vm2 vm3" main_vm=vm1 master_images_clone="ubuntu-18.04-ppc64le" create_vm_libvirt=yes kill_vm_libvirt=yes os_variant=centos7.0 env_cleanup=yes backup_image_before_testing=no restore_image_after_testing=no mem=8096 close_fds=True vcpu_cores=4 smp=4 --vt-guest-os RHEL.7.6alt.ppc64le

JOB ID     : 3b14e850e902735694e0547d33ae3d55cf3a6c40
JOB LOG    : /home/srikanth/avocado-fvt-wrapper/results/job-2019-08-19T07.23-3b14e85/job.log
 (1/1) type_specific.powerkvm-libvirt.multivm_loadstress: /nohup: redirecting stderr to stdout
PASS (301.55 s)
RESULTS    : PASS 1 | ERROR 0 | FAIL 0 | SKIP 0 | WARN 0 | INTERRUPT 0 | CANCEL 0
JOB TIME   : 306.88 s
JOB HTML   : /home/srikanth/avocado-fvt-wrapper/results/job-2019-08-19T07.23-3b14e85/results.html

Where vm1 and vm2 are running uperf workload [server and client], vm3 is running stress-ng workload which is standalone.