You demonstrate sameness of the finished product implicitly by proving sameness of the process
Yes I could see that would work. But how do you demonstrate sameness of the process given that the specific process used is presumably a Teva trade secret? I'm assuming you can't conclude that the processes are the same simply because some finite set of tests can't distinguish the end product of the two processes - that would be circular.
(I'm not altogether clear the extent to which trade secrets and patents can co-exist, and maybe that plays into the picture too).