After updating Zabbix server from 7.0.16 to 7.0.21  we are experiencing strange behaviours in Network Discovery logic on hosts with multiple IP addresses

Steps to reproduce:

1. Create a host with its primary address on its agent interface (i.e. 192.168.64.5)
2. Configure a Network discovery rule with two agent checks:   system.hostname and system.uname. Set scan update interval to 2m. Set Device unique criteria to "system.hostname"
3. Let Zabbix discover the first IP and associate to the monitored host
4. Add a secondary address to the host  (192.168.64.10) and let Zabbix discover this second IP too
5. See Discovery result

Result:

• the first IP is marked as not discovered anymore (with red background, which is WRONG)
• the uptime/downtime counter gets reset every 2 minutes for both addresses (which again is WRONG)
• See screenshot{}

Expected:
The two IP addresses are both reported as discovered, as in older Zabbix release.

This behaviour severly breaks our logic discovery logic on clusters and several other systems with multiple IPs, leading to undesired host deletion. 

I suspect this can be a side effect of the ZBX-26863 fix.