vfio/type1: Remove Fine Grained Superpages detection

vfio/type1: Remove Fine Grained Superpages detection

VFIO is looking to enable an optimization where it can rely on a fast
unmap operation that returned the size of a larger IOPTE.

Due to how the test was constructed this would only ever succeed on the
AMDv1 page table that supported an 8k contiguous size. Nothing else
supports this.

Alex says the performance win was fairly minor, so lets remove this
code. Always use iommu_iova_to_phys() to extent contiguous pages.

Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
Tested-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/0-v2-97fa1da8d983+412-vfio_fgsp_jgg@nvidia.com
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>