Seems you always could "upgrade" to the 'botch' fix found in A7 (It *should* work, as our TPE engineers found it worth to implement in A7 ;-):

a) place some decent 100uF (or as high cap as feasible) parallel to C4110 and C4111. Make sure you don't create another contamination path for RF creeping inside big can - maybe use a ferrite ring to block RF next hole where you route some wires from those big caps inside can to C4110/11. b) *short* C4110 / C4111, and insert the 100uF somewhere else in audiopath, b1)e.g. near R4405 / R4407 (you can even replace these R by 100uF capacitors). You should remove R4117 and R4117 then, and instead place these 1k from JK4401:pin2 and pin3 (! for JACK_INSERT function) to GND b2) or even place the 100uF outside FR in the housing of your hs-plug or 2.5->3.5mm adapter