Yes exactly, that is my thinking, I see no reason from an Azure perspective why they can't be in the same vnet.
With the Azure UDR (dynamically updated by the Azure Function based upon the availability of the primary instance) we would specify the IP address of the active vMX.
It was the first diagram in the guide that was causing me to question this, and think there might be something intrinsic about the vMXs that means that the 2 of them (in an HA pair) couldn't share the same VNet.