On a vSAN Stretched Cluster should an entire Site Fail, ALL VMs will still be accessible on the secondary Site.
vSAN will not rebuild the failed components from the Failed Site onto the Site that stayed up and running.
Remember one of teh golden rules of vSAN> vSAN will never place 2 components from the same object in the same Host/Disk/Disk Group/Fault Domain. This would violate the availability rules for your VMs.
Regarding question nr 2: If the secondary site is already down and if you one single disk/host fails on the remaining site, some VMs will be Unavailable. You cluster is running on a "Reduced Availability" mode, which means that it cannot tolerate anymore failure without causing some service disruption.
The extension of the disruption will be mostly dependent on the type of failure: A failed disk with low number of components is likely to affect less than an entire host failure.