That's an interesting option.
The way the UPX spur is built, any attempt to enhance it as mainline rail service results in a dilemma. Either run more frequent trains, but keep them short to fit the UPX spur design, and thus waste track time on the Georgetown mainline. Or, completely rebuilt the spur at a great cost and with much disruption, to accommodate properly long trains.
On the other hand, light rail should be able to use the spur, after some minor modifications. The direct connection from the terminals to Union would be lost, but the balance of train capacity and demand found. Short light rail trains between the airport and Woodbine Stn, and long mainline RER trains connecting much of Etobicoke and Brampton to Union.
If the same spur can be used to get Finch LRT trains to the airport, all the better. It would be a fairly fast connection, having about 3 intermediate stops between Humber College and the terminals.
Or simply use detachable 4-EMU trainsets, like this:
They come in quick join/detach format so they can scale up/down during peak, or for UPX-specific-route operations.
One 4-train EMU is light enough to go down the spur. The question is whether the bend radius is good enough, and whether the station can be extended to be 1 extra coach long.
Incidentially, this is one of the EMU candidates for GO RER, and it even also comes with dual-level platform door option (high & low level). A mix of low-floor-doors only (for whole RER network) + low+high doors (for RER network *and* UPX platforms *and* UPX spur). Meaning future UPX trains could serve the rest of the GO network as backup trains, and we'd not have a shortage of trainsets available.
Caltrain just ordered dual height door version of the same train:
This is the same train in Metrolinx' clipart (it's only one of the candidates) -- just with the dual-level door option.
I imagine Metrolinx could order 80% of coaches with low-doors only, and 20% of coaches with both levels. Choose "X" percent, replace the numbers, it doesn't matter -- but you get the idea.
That way, all coaches can go on the entire GO network, while a percentage of the coaches are reassignable (at any time) to the UPX network or high-floor platform stations. Massive network flexibility! They'd come in 4-coach pre-joined trainsets, and joinable to become up to 12-coach trains. And currently, the stations on UPX are (just about, barely) extendable to handle 4 coaches -- the station at Union is already 3.5 coaches long when you count ability to meet the door cutouts of the furthest doors (rather than train tip-to-tip) -- 4 coach is currently the proposed Metrolinx standardized length for GO EMUs.
Boarding overhead of bilevels: Although the bilevel nature is subject to debate (read: Sydney, Australia), our network is highly bilevel optimized with great station separation, that we'll more resemble Paris' good bilevel RER experience. I understand the pros/cons, and Sydney versus Paris' differing experience with electric bilevel commuter trains -- but I think given Metrolinx's network design & UPX situation -- it is likely workable without slowing down embarkation too much.
If push *came* to shove, stations can be modified to meet the low doors only (either via raised track or lowered station floors) -- a hugely expensive station modification but that wouldn't be needed right away with dual-level doors -- and you have full operational flexibility of using any GO platform as a backup station while a UPX station is being rebuilt (whether lengthening or platform level change, etc).
It's an imperfect compromise, but has massive operational flexibilities, and permits many paths of future network modifications -- while minimizing mandatory pre-requisites that interact with each other in any phased roll out.
That way, GO RER can merge UPX when electrified.