You've done a terrific job of clarifying the enormous complexity of the assumptions underlying the conflicting views on Keystone XL, without dumbing it down. I'd like to offer a few additional factors for consideration, along with a question and suggestion to clarify the context.
The first factor to consider is the relationship between WTI, which underpins the pricing of most North American crude production, including oil sands, and global oil prices benchmarked to UK Brent. That differential has been extraordinarily wide for most of the last two years but is currently narrowing. Historically it was within $1-2/bbl from either direction (WTI often a premium to Brent) but lately the WTI discount has been as wide as $20/bbl. Today it's under $10. There are good reasons to think that as more Bakken-type crude finds its way onto rail, and as new pipelines and pipeline reversals in the US mid-continent alleviate congestion at the big Cushing, OK hub of WTI, the differential will narrow even further. The closer WTI gets to Brent, the bigger the incentive becomes to find alternative ways for Canada's production to reach market, and the less effective rejection of Keystone XL would be.
I'd also like to put the rail issue into perspective, having spent a fair amount of time looking at and talking about this recently. At a conservative 600 barrels per tank car load (limited by weight for heavy crude, vs. a full 750 bbl for lighter oil) the entire volume of Keystone would equate to 500,000 tank car loads per year. That sounds monumental until you notice that US rail car loads of coal declined by more than that volume from 2011 to 2012. It would roughly double 2012 rail shipments of "petroleum", which includes LPG and other products as well as crude oil.
Finally, I think it's worth asking who is accountable for whatever increase in emissions Keystone might facilitate, and what constitutes the proper basis for comparison. Concerning the former, Canada, as a Kyoto signatory, would be responsible for all "upstream" emissions associated with extraction and upgrading in Alberta. That's more than 20% of the total. The remaining 80% would fall on the US, China or whichever country's refiners ultimately bought and processed the oil for sale as products. In the case of the US, a modest % of those products would be exported, since we're now a net exporter of refined petroleum products, thanks to the convergence of improved fuel efficiency, reduced vehicle miles traveled, and improved refining economics due to the shale or "tight oil" upsurge. You don't have to trace the barrels very far before it becomes clear that this is an international, rather than a US issue. So shouldn't the comparison be on the basis of global emissions, not US emissions? As such, on the worst-case in your analysis, we're talking about at most 0.4% of global emissions of 36 bn tonnes.