Not sure what to tell you. TMX 100 is not materially faster than Delta 100 in Perceptol. Nor is Acros. If a picture of Perceptol as a speed decreasing developer is being painted on APUG, it would be one of the few correct paintings on here
Perceptol, like Microdol, achieves its extra-fine grain characteristics at the expense of speed. It's a tradeoff. One must avoid pushing Perceptol development to full film speed - even when diluted, or graininess quickly increases.
Suggested exposure index and development instructions are usually good starting points, but we don't know exactly how Ilford evaluated TMX in Perceptol and how the measurements were made. In the end, sometimes certain suggested development times, exposure indices etc simply require a little more experimentation on the part of the user. The same can be said of some of Kodak's recommendations. Some of them seem plain wrong. It's nothing new.