No, it seems pretty obvious.
Each use case of processing of personal information must satisfy the GDPR necessary requirement or have a priori informed consent.
During merger of Scribd into LinkenIn, they can argue that it is necessary to treat the databases as a single controller as a cost cutting measure. However, if they are going to treat Scribd as a separate severable entity, then it cannot be argued that it is necessary to process personal information into the Scribd databases which going forward will only process data in a manner not consented to.
If Scribd is not able to distinguish at that point original users that have meaningfully consented to being processed by Scribd from LinkedIn users who have not meaningfully consented to being processed, then they have no meaningful consent at all and therefore no real lawful user data because not knowing if a user has meaningfully consented or not is equivalent to not having meaningful consent for that user.
The most similar scenario is when GDPR took effect and many companies essentially had no meaningful consent for their userbase. During the grace period between GDPR taking effect and GDPR fines being enforceable, companies contacted their userbase to acquire consent and the diligent companies destroyed the user records of the users that did not contact them back since meaningful consent cannot be opt out. Of course, many companies were not diligent and held onto user records nevertheless and are waiting to see if they will have to pay the piper or not.
This is easily distinct from change of a controller when the new controller will continue the business: consent to have data processed in a particular way was acquired. In this case the separated entity will process data in a manner which was not consented to. Unless Scribd is somehow going to try to convince recruiters to use their service to find users who are going to manicure CV like profiles, which is more or less what LinkedIn users consent to.