FWIW I wouldn’t do dish soap like all the time, I only go to it when IPA isn’t doing the job. Come to think of it, since I switched to 99% IPA I haven’t needed to use dish soap at all; doesn’t seem to leave behind that residue.

As always, experimenting is the key. If you’ve eliminated wet filament, z-offset calibration, nozzle clogs, nozzle heat, bed heat, bed cleaning, model issues (e.g., low surface area needing brims), and filament feeding issues… that PLA may just hate you.