It's probably pointless to launch the attack, unless they want to proceed with a 'griefing attack, that screw up network operation. The 51% attack would help improvise enough testnet to generate a great number of blocks in a little time - James Lopp created 165000 blocks in a week using this attack and slowed down the network. The good part of it is that it hurts the functionality of testnet marketplaces, and the delay it causes the workflow of developers who are running their applications on the testnet, is certainly the bad side of the attack, although the action has no benefits for the attacker, it really pisses off developers.