Bug 1769812

Description of problem:
Comparing to previous version (4.2) it takes long time to restore default tuned.

| ver.     | 4.2 |  4.3 |
| new tuned| 31s |  49s |
| restore  |  8s | 464s | its ~60 times slower! 

Version-Release number of selected component (if applicable):
4.3.0-0.nightly-2019-11-02-092336

How reproducible:
100% The best time I had was 86s

Steps to Reproduce:
I was using two terminal windows.

1. Terminal 1 - create cr
oc new-project my-logging-project

oc create -f https://raw.githubusercontent.com/hongkailiu/svt-case-doc/master/files/pod_test.yaml

2. Terminal 2 - check which node will be tuned and debug this node:
oc get pod web -A -o wide ##to get node

oc debug node/<my_node>
# on pod:
chroot /host

i=0; while [[ "$(sysctl kernel.pid_max | cut -d ' ' -f 3)" != "131074" ]]; do sysctl kernel.pid_max; sleep 1;i=$[$i+1]; echo "time: $i"; done

## this will create loop to check changed value by NTO

3. Terminal 1 - label pod and create new tuned.

oc label pod web -n my-logging-project tuned.openshift.io/elasticsearch=

oc create -f https://raw.githubusercontent.com/openshift/svt/master/openshift_tooling/node_tuning_operator/content/tuned-kernel-pid_max.yml

## Creating new tuned check time on Terminal 2

4. Check time when node will be tuned by new tuned.

5. Do something similar deleting tuned.

6. Terminal 2 - check node for changes

i=0; while [[ "$(sysctl kernel.pid_max | cut -d ' ' -f 3)" != "4194304" ]]; do sysctl kernel.pid_max; sleep 1;i=$[$i+1]; echo "time: $i"; done

7. Terminal 1 - delete tuned

oc delete tuned max-pid-test -n openshift-cluster-node-tuning-operator

## deleting tuned check time on terminal 2

8. Check time when default value on node will be restored.


Actual results:
Restoring default values taking too long comparing to previous versions.

Expected results:
Similar time to restore default values.