Even with the current schedule, you need at least 5 tracks just for Lakeshore at peak. For example, there's a departure at 7:40 AM eastbound to Pickering, and then arrivals at both 7:43 and 7:45 from Oakville, 7:43 to Aldershot, and then another 7:50 arrival from Pickering.
Four tracks might work well if there was just off-peak service, but it's the express services at rush hour that make things more complicated.
Agreed; the assignment of express/peak bilevel diesel services requires separate dedicated tracks even in the GO RER era.
I'm just referring only to the baseline allstop electric RER services. Even with electric-locomotive-driven bilevels, with shortened dwell times and electricification, you can probably run 7.5-minute RER frequencies per platform assuming the USRC efficiency is sufficiently improved by then (resignalling, new train traffic control centre, higher-speed crossovers, automatic train control, available corridor capacity after contending with other services, etc). When the same platform has another train coming in 7.5 minutes (ala Sheppard Subway frequency), you only have to open the GOtrain doors for about 1 minute (or a bit less) before closing and accelerating out of the berth right away for the next train. It's doable even with low-platform trains, assuming USRC is optimized.
Consider the potential of 7.5 minute RER peak frequency Lakeshore East/West using electric locomotive driven 12-car Bombardier bilevels, and 7.5 minute RER peak frequency "SmartTrack" using EMUs (see my other posts why I think the Kitchener-Stoufville GO RER will end up using EMUs). With efficient service operation, there is already mostly sufficient corridor capacity on Georgetown corridor (if EMUs) and Lakeshore West (with electric locomotive driven bilevels) after some optimizations. France does higher same-platform frequencies than this with their commuter trains, and we already do ~7.5 minute peak frequencies today on Lakeshore with just diesels albiet interspersed on different tracks/platforms.
7.5 minute peak same-platform frequencies would empty the platform quicker with quicker train turnover, keeping the platforms clear in the era of an expanded Union. Otherwise, platform overcrowding will otherwise become a worse problem when number of peak period passengers double at Union through 2031.
Still, 5 to 7.5 minute peak-period turnover per RER track is possible with low platforms, assuming USRC optimizations are milked to their potential. Though, interspersing with other services on Georgetown corridor will be challenging. With EMU's, you can pull off shorter dwell times, and with raised platforms, even shorter dwell (towards subway frequency), but the USRC throughput capacity will almost certainly be the limiting factor. How much long-term throughput Metrolinx wants to milk per track, will require several very important decisions over the coming years.