Knobs to tune logging for network-operator

[adrianchiris]

What would you like to be added:

a way to configure log level for network operator and possibly its operands.
in a way that will not require a redeploy of the operator.

Why is this needed:

currently operator has zap-devel logging turned on, it is not structured and emits debug logs by default.
in production env we would most likey want INFO level logging as well as logs to be json encoded.

as for network operator operands, it may be a separate mechanism.
cluster disruption should be taken into account, which may mean we will not do it for the operands.
possiblitly: extend nicclusterpolicy per state with logLevel were applicable and let user decide.