enhance the CPU requests calculation when LATENCY_TEST_CPUS is not set. but even if it is set we need a better way to ensure we are working with valid amount OR completelt pour that responsibility on the user. 

all supported versions

    depends on the node load 

    1.
    2.
    3.
    

    * latency pod should be created with cpus request that can fit in the worker node (nodes?)
    * rules for noisy neighbor and other rules for minimal allowed values should be kept
    * we need to consider the case where PP is configured with single NUMA policy and the pod requests CPUs that are only available in node level but not under one numa; in this case logout a warning & skip the test. Another option would be to allow setting the scheduler for the latency pods (a smarter scheduler like the NROP one)

    we need a smarter way to calculate the node base load from other pods before setting the cpus request for the latency pod https://github.com/shajmakh/cluster-node-tuning-operator/blob/master/test/e2e/performanceprofile/functests/4_latency/latency.go#L328