Your specific cause Connection prematurely closed by remote server when sending DNS requests to unbound has been intensively analysed and identified as being caused by specific behaviour of unbound terminating queries if it exhausts its TCP connections, see Two Pi-hole instances, One with "failed to send..." - #44 by DL6ER.
While there is nothing that could be improved on the Pi-hole side, increasing unbound's incoming-num-tcp (e.g. to 40) is reported to mitigate the issue, though it may still reoccur under heavy load at times.
In all likelihood, this error has occurred for Pi-hole v5 all along, but Pi-hole v6 started registering it. If you didn't notice any issues with v5, then this is probably nothing to seriously worry about.