Possible, but hard.
Let's use some common sense here...
The engineering involved would rival that of Toyota's (pretty complicated stuff).
You'd still need an IACV, EGBV (or their equivalents) and several VSVs to control them connected to a stand-alone ECU with additional outputs.
I have no idea whether or not the #2 will cause the #1 to spin backwards, but common sense dictates that there would be a loss in efficiency for both, since the #1 is being retarded (due to drop in the pressure difference between the charge air and exhaust side) and the #2 is working against #1, but either way you'd have to gradually bring #2 up instead of slam-starting it, again just like the stock sequential system.
Maao, the "direction" of the air doesn't matter... as you learned in high school chemistry, pressure equalizes throughput a closed system and does not flow like a fluid does downhill. The air doesn't know where to go; it goes everywhere. In a car, it just so happens that the pressure differential between the turbos and the engine (engine creates vacuum by consuming air and burning it) causes the air to be sucked into the manifold and into the cylinders.
There's always going to be static pressure in the manifold (positive boost), so you'd probably have to shut down the #1 turbo once the #2 comes on (as opposed to the sequential, then parallel, operation of the stock twins). Basically you'd replace #2 with a turbo that has greater output than the stock twins combined and you'd come out on top.
Bottom line; the two turbochargers would have to be segrated using a system very similar to the stock sequential to avoid complications due to boost difference. You'd probably need 2 separate wastegates as well, connected to the system that controls boost management. But once #2 comes on, #1's wastegate will have to be closed to use all the exhaust pressure possible to power #2, and the manifold going to the #1 will have to be blocked off to prevent unwanted complications. You'd probably still have a pressure drop in the transition period.
Yeah, probably $20,000-30,000 to engineer, not to mention the test equipment. That's why companies put R&D into complicated turbo systems and not individuals.
