You asked a narrow question to which there is plausible (albeit unlikely) cause(s) of action.
Now you assert that MNTA/Sandoz (MS) "should" have no idea whether TEVA did infringe.
Maybe - maybe not.
It is plausible that accomplishing the task (characterization, producing, proving "sameness) without the use of the patented processes is so difficult or costly that MS can infer infringement. That is sufficient for suit. More will be required to win. To win there will need to be proof of direct infringement or under the doctrine of equivalents*.
Discovery will be the means of obtaining the proof - if such exists.
Market Data 
Markets 
AI
