That makes sense, and would be true if arrivals at the traffic light were random. But in my experience the streetcar is almost always sitting there before the light changes. It would be interesting to go through real data though.
If arrivals at the traffic signal are random and the light is red 60% of the time (which is the case at major intersections along Spadina), then 60% of the streetcars will be sitting there before the light changes.
All downtown signals operate on a fixed cycle length (except during TSP calls), so if you want to change the offset from one signal to another, you change that by changing the programmed 'offset' value. No need to change the phase order.
In general, signal coordination wouldn't work well for streetcars if there are stops within the block, since variable dwell times basically randomize the arrivals at the next signal. It's also worth noting that coordination typically only works in one direction, not both.
Nevertheless, there are some cases where signal coordination can be used to provide better-than-random offsets for streetcars. For example during the King Pilot there was a westbound green wave from Simcoe to John, since there are no westbound stops in that block. I observed the John St intersection for an hour one afternoon, and every westbound streetcar with a headway over 90 seconds got a green light, thanks to a combination of coordination and signal priority. There was also an eastbound green wave from York to Bay, but since there's no TSP at King & Bay, streetcars could still get stopped at Bay if the operator drove slowly or there was some kind of disruption between the intersections (e.g. a car parking).
I also requested to introduce a green wave for streetcars on Cherry Street, since there are blocks with no stops and the intersection spacing and cycle lengths happened to be ideal for a perfect green wave for streetcars in both directions. But the City denied the request on the basis that it would make the coordination worse for cars. As far as I'm aware, the King pilot was the only time the City has specifically optimized coordination for streetcars rather than cars.
Though by putting the streetcar phase first, there'd be less issues with cars still leaving the intersection and it would be easier to predict when you'll need the light green.
By that logic, putting the streetcar phase first would make the situation even worse, since the previous phase would be the cross-street phase, which lets far more cars into the intersection than a left turn phase. It would also make it harder to provide a green at the right time since it's a lot easier to end a left turn phase than a thru phase, since left turn phases don't have a Flashing Don't Walk.
A better idea - if you put in a real intelligent system, why not let it choose either before or after in realtime, based on what works best for streetcars on that particular cycle.
That's known as phase rotation and the TTC's current priority system can apparently do that, but it has never been approved by the City. The problem with phase insertion is that in a cycle where the thru phase comes before the left turns, it serves its full (quite long) duration, so there's a very long period where no left turn phase is served. That could cause the left turn queue to spill back into the thru lane, strangling the intersection throughput. Especially where the side-street is relatively minor, it's less of a nuisance to just serve a short streetcar phase before going to the left turns, rather than the full thru phase duration. And phase insertion works just as well for streetcars.
