Modern CPUs are extremely capable, but if concurrent in-process commercial detection is a concern, it could be limited to only run against 1 recording at a time, and perhaps only for the one that is actually being watched. Or batched and queued. Comskip doesn't run continuously. Process one recording, then the other. If the first accumulates more in-process recording, the run it against that unprocessed portion once other tasks are complete. This is not an unsolvable problem.
Why do optional steps belong only in post processing? I don't see the logic behind that. If it's an optional step, the optionally run it after a recording is complete, or optionally run it as the recording is in process. Segments of video that are stored on the server can be batched and queued regardless of being in-process or completed recordings, with identical resource implications. They are being processed one-at-a-time regardless.
You're right, I do want the missing aspects of SageTVs feature set in Channels, and I do run both SageTV and Channels concurrently. SageTV has other architectural limitations that prevent it from offering everything that Channels does, such as broad client compatibility. I don't understand why you would take the stance that Channels shouldn't implement the valuable features of other platforms. It was Steve Jobs who said “We have always been shameless about stealing great ideas.” I suggest you also give SageTV a try, understand it's limitations, and also experience for yourself some of the great features it offers. If you actually try in-process commercial detection, you may find you actually like it. Or maybe you don't watch sports, and don't care about skipping commercials and being caught up to real time at the end of the game, and that's OK too. Not every use case is for every user.
For your suggestion that I pause on a commercial, go do something else, and then return and fast-forward, I think that's a terrible user experience. Why should the user have to change their behavior because of a technical limitation of a product? The goal of a product should be to solve the problems of the user, not to make the user behave differently and put up with with the technology.
And to be clear, these are all "first world problems". This isn't about curing cancer or ending world hunger. It's about making a perfect user experience where the technology solves to problems of the user. And the problem I wish Channels would solve for me is, when I sit down to watch a football game 30 minutes after it starts, skip all of the commercials automatically so that I'm caught up with real time by the end of the game.