I don't think aspects of "good" or "evil" should ever be assigned to an inanimate object. Its potential use by an individual for good or bad purposes and how such use is perceived as either good or bad by others makes it moot.
Sure, the programs running within the holodeck can "act" good or evil based on its programming, but again, that programming came from elsewhere outside the holodeck. I guess this runs back into the whole "is a holo-person really sentient" question, but I would say no.
I wouldn't even classify Moriarty or holo-Leah Brams sentient, as that would imply sentience on the part of the Enterprise computer systems itself, as the controller of the algorithm. If THAT were true, it would be a very different kettle of fish, but the computer never exhibited signs of sentience before or after that time and, if it did, Deanna Troi would have had to go to trial for murder for crashing it in Generations!