@the_rock gave info on how you can do this with the Endpoint VPN client, but for site-to-site VPNs with 3rd party gateways, this can't be done with pre-shared keys.  If the remote end is willing to do either:
- Use a certificate issued by your management internal_ca, or
- Use VTI with IKEv2 and universal tunnels  (likely the better choice)
 
With certificate, the remote end can allow your side to appear as a "dynamic IP" VPN peer and allow authentication to rely on the certificate. However, now that we have VTIs and universal tunnels, certificates will be less preferred and instead you should use the VTIs.
The remote end will build 2 VPN gateways for each of your ISP-provided IPs (or the cluster VIP for each, if you have ISP Redundancy on a cluster).  The encryption domains for both you and them will be empty group objects (or, however the remote device does this).  You'll then create a VTI on your firewall gateway in CLISH, as will the remote peer.  The IPs on the VTIs are irrelevant; VTIs are just point-to-point links and any packet you send to that VTI device will be encrypted and sent to the specified remote peer (specified in the VTI configuration).
You can use static routing with either primary/secondary priority, or just let it be equal-cost; doesn't matter either way.  This is how you do an AWS VPNgw redundant VPN, too.
The remote peer does the same thing, only in reverse, with a VTI on their side.  They will add static route for your VPN networks across the VTI.  Configure the VPN community tunnel management as "one tunnel per gateway pair" and you're done.