I think this discussion may be getting rather confused/confusing ...And I'm thinking of a fitting where the seals aren't quite right letting in gradually more moisture, or a capacitor failing gradually. .... So I think you need both a timing difference (for the fast faults) and a setpoint difference (for the slow faults) to achive effective discrimination.
Either a standard or TD RCD will only trip if the residual current exceeds its trip threshold, in both cases regardless of whether the residual current has reached that threshold very suddenly or after a long period of gradual increase. The only difference is that the tripping process will commence immediately after the threshold has been crossed in the case of a standard RCD, but will not commence unless/until the residual current (above the threshold) has persisted for a certain ('delay') period of time.
I can't see how the 'history' of development of the fault/residual current (i.e. how long it was rising for prior to passing the trip threshold) can have any bearing on the behaviour of an RCD.
Of course, if you want 'discrimination' in the sense that one device will operate earlier in response to a gradually increasing residual current, then the two devices obviously would have to have different trip thresholds, but that's a totally different matter - and not what we are discussing.
That's how I see it, anyway!
Kind Regards, John