Jamil Zainashef recommends pitching onto a yeast cake in his brewcast episode for either a maibock or a dopplebock, I forget which. He's won a competition here or there, oh, and he has that pitching rate calculator thing. But you're right, its NEVER a good idea to pitch on a cake.
Trying to decide if that's a passive-aggressive snark or not. I'll go with my gut and say, "Not."
First, Jamil is not infallible. While he's not technically wrong, some of us have more brewing background than homebrew competition wins.
Second, even with high-gravity lager beer, I can't think of a situation in which the dregs from a 5-gallon batch
won't be overpitching - unless the fresh wort is
10 gallons of 1.100+ bock wort. Then, yeah, you might have a point.
The Rule of Thumb for pitching is 1 million active cells per milliliter of wort per degree Plato. Thus, for a wort of OG 1.048 (12°P), you need 12 million cells per milliliter of wort. There are 3785 milliliters per US gallon, or 18,925 per 5-gallon batch. Thus, you need 227.1 billion cells to properly inoculate the wort.
Fix et al. estimate approximately 1 billion yeast cells in 1 milliliter of slurry, due to trub and other non-yeast material (that number is massaged slightly based on personal experience). That number goes up when the slurry is washed - not important for the present example, but keep that in mind. Thus:
227 ml of slurry is required.
You'll forgive me if I consider an ale pail with 2" of slurry on the bottom contains just
slightly more than 227 ml.
Even when considering a wort of 1.097 (23°P), you need ~435 ml of slurry. That's still about half of what I get in the bottom of an ale pail after a "normal" ferment.
So yeah, if you're brewing freaking gigantic beers, or doubling the batch size (brew length)
and brewing a bigger beer, using all the slurry makes sense.
Still doesn't stop my "Look at that filthy fermenter OMGWTFBBQ!!!!11!!" button from being pushed, but I can't force anyone to be clean.
Cheers!
Bob