Responses are queued to be sent in the next second. This restriction makes sure that a VSE running on a client-sized PC doesn't overload that PC, regardless of whether it's a laptop, a virtual machine, a container, etc.
If you want to do performance testing against a VSE, or you have a critical need to always have responses sent with the exact delay timings you specify, you want to performance-tune a machine on which to execute VSE and implement a "Performance VSE" component instead of a regular VSE component.