I'd say with the advent of the Townsend Lab Sphere microphone, component modeling and the incredible technology behind the Acoustica Audio (voltara kernals sp? - more capture than modelling) there are no significant gaps in analog modelling technology today.
A big part of the appeal of analog gear (to the proponents) is the commitment you make to setting levels and tweaking with knobs. You have a hands on experience with it and we form attachments that way. We also consider those "hand" created sounds special. The sonics may not be any better but the process in getting there may mean something substantitive.
That is not say you can't do the same thing with any softsynth. YOu spend time with it and manipulate it and come up with something that is yours and the attachment is the same. It is simply that touch is a powerful sense and analog typically requires more of it by design or limitation (you choose).
If a small degree of non linear distortion (whereby an oscilloscope is needed to detect the difference between analog and emulated), is a make or break deal for the track, it must be pretty unremarkable bit music.
Asusp9712Core,RayDats,OASYSPCI,Cubase9,KorgLeg,Avalon737s+2022,UA4710D,6176,DCS,LA610MK2s,Neve, ARC2, Soundtoys,UADOCTO,AdamA7X, SphereMics, BCF+R2000s, DrmBrut,MiniLog,Volcas,MoogSubPhat,EnsqDP4,VirusC,Scope,MinBrut,Plugiator,NIS61 Ultimate,ComposrsCld.