Electric railway guy here.
Long distance propagation
I have seen 600V trolley wire dip to only 200V four miles from the substation under heavy ~300A load from a single articulated car. (4/0 wire, 107 mm2, rails as return).
Third rails are a great deal beefier, but subway trains are a great deal heavier. Typically third rail shoes are fused at 400 amps (per shoe, and not every shoe is in contact at once) with as many as 8 cars. Oslo runs big articulated cars that are electrically 3 cars.
If the regenerated electricity passes a substation, it's even more at disadvantage.
I mean the subway train could push its regenerated power any distance if it's willing or able to increase voltage without limit. Unregulated, DC motor regen can act like an old, inductive constant-current source, increasing voltage until current flows. Burning up much of it in transmission losses would be fine, it's "free energy". However it hits limits of a) onboard equipment (not least, insulation strength in motors), and b) the third rail. BART aimed to have a 1000 volt third rail, but found the worst case scenario of rain on brake dust caused spectacular flash-overs even in their temperate climate. They backed down to 900 volts but it is still troublesome. Oslo is already at 750, not much headroom.
Really, to regenerate productively, there needs to be a train nearby already pulling the voltage down and able to gobble up those amps.
Regen onto grid
This is hard, not least because a couple megawatts of power injected for a few seconds isn't all that useful to the grid.
Also, DC-AC regen itself is hard, with large silicon inverters required at every substation.
In the Golden Age, rotary converters were perfectly capable of efficient DC-AC regen (in fact, they had circuits to prevent accidental regen, e.g. a substation's local grid having a brownout, causing it to be backfed from another substation via the trolley wire). Electric railways had more of their own AC power distribution. And third rail voltage was only 600V, so more headroom. However, the cars were not capable of it: subway trains were very simple back then, with only 7-12 wires on the inter-car control lines.
Rotary converters were abolished just as soon as mercury-arc rectifiers became available, and even those were gone by the time of the first regen cars.
I don't expect any resurgence in rotary converters (more's the pity, since they are dog simple, actually correct power factor in the local grid, and may be competitive since they are simple). So it comes down to complex, large inverters. Given the limited financial gain from selling power back, only very advanced (high R&D) systems like BART are dipping their toes into grid regen from DC.