Z-wave is handled via mqtt these days, so you don’t need the added complication of two HA instances. Just set up zwavejs2mqtt on the pi, and use the mqtt broker on your ha instance.
FWIW, I have a master-slave setup via MQTT.
The point is to put radio related stuff (zigbee-ZHA, BT) as far as possible of interference sources (mainly the USB3/SSD ones generated by my main HA setup on RPi4)
As those two are basically incompatible (go figure), I created a blueprint that publishes discovery-compatible topics based upon the statestream one: MQTT State Stream to MQTT Discovery
Or, I believe you could use it with the preferred websocket integration. zwavejs2mqtt also supports that and it is the much cleaner 2 way communication channel used by the HaOS addon.
With core I didn’t reference the install method, but the integration / Add-on being maintained by the core team, opposed to custom or community add-ons
OP asked for users to share their experience on easiest ways to get them to talk together. So I did share mine
Hey all… quick question on Remote Home Assistant… my use case is to have a second RPi running all of my cameras, and have the master running all of the devices and automations… the reason is to reduce the CPU burden on the main RPi by managing all of the camera feeds… so if I run a multiple instance like suggested above using Remote Home Assistant, with cameras on the slave and everything else on the master… will the slave essentially still burden the master with all of the camera feeds, or will the slave maintain the burden and yet allow the master to access the entities of the camera triggers ?