Questions on R-Series 5800s

For the LAG question, yes 
on rSeries a LAG can span ports from different pipelines as long as the member links are configured with the same speed and type. So in your example, using 3.0 and 7.0 in the same LAG should work if both ports are configured identically. I’ve seen deployments where links from different pipelines are used without issue, but the usual recommendation is still to keep link characteristics consistent.

For tenant deployment, I usually start with "Recommended" unless there is a specific reason to manually tune resources. Advanced provisioning is mainly useful when you want to customize memory instead of using the default vCPU-to-memory ratio

For virtual disk size, leaving the default is generally fine to start with. Tenant disks are thin-provisioned, so the configured size does not immediately consume that amount of physical storage.

If you want to adjust vCPU or memory later, it can be changed, but the tenant must be moved to the Provisioned state and then deployed again. Depending on your setup (HA configuration, traffic mirroring, etc.), this may have service impact during the redeploy.

References:
https://clouddocs.f5.com/training/community/rseries-training/html/rseries_networking.html
https://clouddocs.f5.com/training/community/rseries-training/html/rseries_multitenancy.html