When decommissioning a failover, we get the option to DELETE or DISABLE, but what is the difference for the backup agent actually? According to the info on the popup the result would be the same.
Deleting the failover config will stop the backup agent. The DataMiner ID and all data will be removed from the backup agent. This agent will no longer be able to start up. It's completely reset (you can compare it to a factory reset). This is done to prevent double polling (2 agents polling the same devices). The online agent will release its virtual IP and inform the rest of the DMS that it is now reachable through its actual IP (and no longer the virtual IP).
Disabling the failover config will not delete data, nor the DataMiner ID, the configuration will remain present but will not be "active". The online agent will still release the virtual IP and inform the rest of the cluster that it is no longer reachable through the virtual IP.
For failovers running Cassandra, their Cassandra cluster database will always be reverted to single nodes. This is done on both delete and disable failover config.
The main difference between disable/delete is that you can enable the failover configuration again after disabling it, while you would have to reconfigure it entirely after deleting it.
From the image in your question, it looks like the popup is not completely correct. The backup (or offline in the screenshot) agent will no longer be reverted to a clean installation when failover config is disabled since DataMiner 10.0.11.0. I'd suggest contacting development to fix this in the software.