This is the long version of what I settled on in the end and why:
I read up on o ring splicing and talked to some friends, and apparently it's common engineering practice to use a glue similar to superglue to splice o rings.
Permatex actually makes a specific o ring glue you can buy and lots of places sell lengths of straight rubber to make your own o rings. Since the o ring did have some use on it, and the cost of the glue and rubber to make a new o ring is more than the cost of an actual o ring, I decided not to join a new piece in.
I went to Canadian tire's plumbing section afterwards, and there happened to be an o ring the exact same diameter in thickness, and just slightly(i.e. 1-2 mm) too short to fit. Of course once you tighten the cases together the rubber expands somewhat anyway, so I laid the o ring in the recession and placed a wooden block on it for about a minute. Removed the block and it had taken exactly the same shape as the old o ring minus wear on the old one of course.
In general you shouldn't stretch an o ring, but if you stretch it less than it will expand where it's seated it will be fine(advice from a mechanical engineer friend).
Another reason not to use the superglue in this position is that I don't know how well it stands up to solvents like gas, and since it's a water pump I didn't know if the heat would exceed the 130 degrees celsius the super glue is rated to withstand.
I did however test a tube of *new* superglue on a spare o ring. Once cut and joined with the glue it was almost impossible to break the join. Typically pulling on it, it would tear just slightly after the join, indicating that the superglue was plenty strong, but the rigidity of it caused the nearby rubber to absorb more strain in situations where tension is placed on the o ring. To the guy who had an o ring fall apart in the carb, I suspect gasoline destroyed the glue, the permatex o ring glue may be unaffected by solvents, but I didn't look too far into it.