Enable TW pruning, no remapping

[PaulZhang12]

Summary: Enable TW pruning for TorchRec inference modules. We switch from pruned_indices_remapping to num_rows_post_pruning given the fact that the indices remapping isn't calculated until after physical transformations. Logical transformations (optimizing the model itself) should not depend on the reset of physical transformations, and the actual index remapping will be set later on in model loading

Differential Revision: D61879687