To the second part of the question, there are a number of apps that use smart trainers. You use two. Training-oriented apps like Trainer Road or Sufferfest or virtual worlds like Zwift, Rouvy, and RGT take your power from a smart trainer or an on-bike power meter, or an estimate of your power from a trainer's resistance curve. The former class of app is mainly interested in your power, although it may control a smart trainer's resistance to simulate climbs where you need lower cadence, high-resistance efforts, or to do ramp tests (which estimate your power around your VO2max, and then estimate your functional threshold power from there; these tests involve ramping up the power until you can't pedal anymore). Virtual world apps move you through a virtual world, and they will vary smart trainer resistance to account for terrain.
I don't use Peloton. My understanding of Peloton bikes, based in part on this article by DC Rainmaker, is that they do measure your power, but they do not broadcast it over ANT+ or Bluetooth, which are the two open source protocols that just about all cycling devices use. Additionally, Peloton bikes have controllable resistance, but because they don't use the two transmission protocols I mentioned, a different app can't control them. So, a Peloton bike will be inherently less flexible. That said, someone is using Kickstarter to fund a device that could render a Peloton bike controllable via Bluetooth. (In general, we don't give product recommendations, so this isn't a purchase recommendation. Additionally, not all Kickstarter projects succeed, and you usually get left empty handed. Or they could succeed with a lot of delays.) If this device or similar ones came to market, this could render Peloton bikes more flexible. Peloton doesn't seem to have an incentive to re-engineer its hardware to broadcast over the common transmission protocols.
Approaching the question from the other side, we could ask what inputs Peloton classes take, or perhaps what inputs the standalone Peloton app takes. I am not certain. It does appear that you can use the app on any spin bike or even a real bike (although presumably not on the road). Mom Tech Blog seems to imply that the Peloton app can be run without any sensor input, e.g. heart rate or cadence, at all. The app may not be designed to acquire sensor input and transmit it to the instructor. Sipping and Shopping, another blogger, described how workouts often rely on a certain cadence, and she got a cadence and speed sensor for her spin bike. She used them with the Wahoo app. However, again, I'm not sure if your cadence from a Bluetooth cadence sensor would be transmitted to the instructor. You might not get personal call outs from the instructor this way. This might or might not impair the experience enough for your wife to be dissatisfied. It does appear that the app may have a short free trial period, as do all the apps I mentioned in the first paragraph. Also, you can manually change the resistance on your smart trainer from its own app, so this might help you stay in tune with a Peloton session better than a regular wheel-on trainer.
I'd speculate that Peloton is aiming at a different market than the cycling-oriented apps. Peloton might never really feel the incentive to integrate external sensor input into their own app. Many dedicated cyclists would probably weigh the cost of the Peloton bike against a road, mountain, or gravel bike, a premium set of carbon wheels, or something else. That said, I don't know Peloton's breakdown of profits and losses by segment, but I would guess that the subscription part of their app may be more profitable than the bikes. Bikes, after all, are mechanically complex and require service and maintenance. It's possible they might decide that modifying their app to take external sensor inputs could be a potentially profitable move, or at least a relatively low-risk one. Feel free to prod them on social media.
In terms of live classes, I know one cycling studio near where I live. (NB: I'm linking them as an example, not as a recommendation.) Their business model has centered on in-person indoor training, coaching, bike fitting, and other services. I don't think this business model is that common, but I suspect it's not unheard of, at least in areas with a big cycling community. This store has done a lot of online classes during the pandemic. It's possible this segment of the market will grow and will offer live but remote classes for people who don't want to be there in person. The rationale for this is that the indoor segment is much more conducive to structured training. For instance, I don't believe I have a segment of uninterrupted road or bike trail where I can reliably perform a 2x8 minute FTP test - I can get there very early in the morning such that almost nobody is present, but there's simply not enough road. For competitive but time-constrained athletes, structured training can be more time-efficient than unstructured training.