Strangely enough, Server version 7.2.1 was not publicly released. Perhaps do you refer to 7.2.0 or 7.2.2?
There is no factory limitation of one update every three seconds, so the limitation can be in the source or perhaps in the communication.
We need to see a snippet of the Server log covering a whole client session which experiences the issue.

Can you replicate the behavior with a test installation of the Server? In this case you can enable the following loggers:
- LightstreamerLogger.pump at DEBUG level
- LightstreamerLogger.subscriptions.upd at DEBUG level
- LightstreamerLogger.requests at INFO level
and we can trace the lifecycle of the updates.

In case it only occurs in a production installation, can you check if the Server process is overloaded?
Alternatively, you can collect the log with factory setting but adding the configuration of the LightstreamerMonitorText logger at TRACE level (which means one monitor line every two seconds).