Does anyone remember issue of transaction malleability and quadratic sighash before SegWit exist?
Yeah, and it is still present, because you can still use legacy addresses. It could be fully eliminated, if legacy addresses would be obsolete, would be fully unsafe, and as easy to move as OP_TRUE is, and then, they could be made non-standard (not "invalid", because of the past timelocked transactions; as long as there are any legacy UTXOs, they cannot be fully invalidated).