We will be using the Common tenant for this (first), as it is best practice to place resources within this tenant if they are to be shared across multiple tenants. Using the Common tenant is not the only way, though; we can also use the Fabric Access policies to achieve the same goal for the Infrastructure tenant.
Why would you use one tenant over the other?
Using the Common tenant means that the DHCP relays can be used by any tenant. If we use the Infrastructure tenant, the DHCP relay policies are selectively exposed to the other tenants. Configuring a global DHCP relay under the Fabric Access means that any tenant can use the DHCP relays and we have a higher level of control.