You already have tons and tons af recordings of words built it and organized with an entire database. How about adding some sort of add-on that systematically goes about using your huge database of recordings to do general purpose chinese tone training?
if you set your flashcards module to "Test type" Self-graded and "Show" Audio, you can already do something similar. Pleco will play back the syllables to you, and you will have to recognize both the right tone and pinyin and, if desired, remember the Hanzi with it. You could also create a category with only single-character words for use as your test set, to make sure you don't get any clues from word combinations.
yeah... I could make a deck for it. But I wish there was a feature that made use of the already extensive set of word recordings and dictionary to provide an extensive guided automatic stinoslice for learning to recognize tones well.
Honestly I don't know if that would teach tone recognition that well - words in isolation aren't remotely the same thing as real speech, it's way way easier to pick out whether somebody speaking a single syllable or word very slowly and deliberately is using a 2nd or 3rd tone than to tell the difference between those in actual native speech.
If we wanted to build a feature around tone recognition we'd probably be better off licensing a bunch of high quality recordings of longer-form spoken audio and testing you on bits and pieces of those.