The malleability of the ciphertext matters because it enables certain circuit tagging attacks as the article explains. It means that the exit relay could confirm you are using a guard relay also controlled by them and thus discover your origin IP address.
You can indeed use HTTPS with the end server (e.g., accessing Wikipedia). This correctly hides the traffic content from all relays.
To reach this point, though, you first need to set up the Tor circuit itself. This is done in a 'telescopic' fashion: the user connects via TLS to the first relay, then sends a message to extend the circuit to a second relay, then to the third (and usually last) relay. Finally, to open Wikipedia, you send a layered encrypted message to the last relay. All this data is link-protected by TLS on the wire, but protected by Tor's relay encryption mechanism while being processed by the nodes.
There are many reasons that these cryptographic tagging attacks are a lot worse than just the timing correlation attacks that are possible if you control the guard and exit of a client: https://archive.torproject.org/websites/lists.torproject.org...