If all 4 cylinders dump into one large collector volume, you lose a bit of kinetic energy.
So the "divided scroll" or "twin entry" idea is kind of like this - take the turbine housing, divide it in half, port 1 & 4 into one half, and 2 & 3 into the other half, then the transition from "off boost" to "boost threshold" should be aided by the maintenance of kinetic energy contained in the syncopated gas flow between the pulse-timed cylinder pairs. It all sounds good in theory, but if you make a convoluted manifold that sacrifices gas flow for pulse-tuning you could very easily give away more than you gain. Within the confines of a MR2 engine bay, it might be difficult to properly execute a decent header for a divided scroll turbo. Pictures of the old turbocharged F1 cars with 4-cyl motors clearly show the application of the divided-scroll theory applied to reality, so I'm guessing there is some merit to it. A prominent bord member who is in the business of making headers claimed the divided-scroll method does not give you any measurable benefit, although I think the benefit is more in transitional presure increase vs. RPM, not in peak horsepower numbers in an almost static mode.
Either way, it's a matter of opinion/choice. Do you choose to make a very complicated header, and spend extra money on an expensive turbine housing, or just go with the flow (no pun intended) and run a regular turbo?!? We are talking about changes in boost threshold of maybe 300-500rpm at best, and admittedly, most of this technology and approach is from the era of turbos BEFORE the advent of twin ball bearings, and ceramic ball beraings and things of that nature, so which is best?
standard old turbo?
divided-scroll turbo?
single ceramic bearing Turbonetics turbo?
twin ball bearing Garrett turbo?
twin ball bearing Garrett turbo with a divided-scroll?
