Varnish IP's staleness should not be checked in request handler

[GabrSaac]

When the current time is after "resetAfter" hundreds of requests could call "getPrivateIPs" because "resetAfter" is set to a new value only when the "getPrivateIPs" request is fulfilled. (resetAfter can also incur in race condintions - see issue #2)

`

Requests handler (concurrent) :

privateIPs := taggedInstances
   // Check instance cache
   if time.Now().After(resetAfter) {
    privateIPs = getPrivateIPs(ec2region)
    resetAfter = time.Now().Add(time.Duration(*cache*1000) * time.Millisecond)
    taggedInstances = privateIPs
   }