At first I thought that was incredibly wasteful, then I realized it was actually brilliant, given the circumstances in which the plane was used.
The pilot told the Crew Chief via intercom "Engage Buicks". The Buick operator pressed "Jet Start", and movement of the Buick throttle (one throttle lever controls both engines) downward engaged the transmissions. . . . The pilot would watch his onboard gauges to confirm minimum oil pressure, fuel pressure, and rising RPM, and then set the jet throttle to idle. At that point 30cc of TEB was injected into the burner cans, a characteristic green flame was emitted, the J58 lit off, accelerated and started to run on its own. The accelerating jet started to unload the Buicks, and when 3,200 rpm was reached, the pilot called "Buicks out", the Crew Chief signaled "cut", the Buick operator hit "Cart Shutdown", the probe fell free passively, and the cart throttle automatically returned to idle.
https://www.thesr71blackbird.com/Aircraft/Engines/starting-t...
So the gist of the idea is not that far fetched.
(Now I'm wondering how the huge Diesel engines on massive boats are being started, I wouldn't be surprised if they were started by a smaller IC engine)
I’m totally speculating here.
I wasn't there but I have the impression that kerosene was not the limiting factor back then. So they solved an engineering problem (body expanding in multiple directions) by declaring, not a problem!
But then again, maybe while designing you observe that there's going to be a problem. You do realize that, say, the very different temperatures will inevitably produce too big an expansion for the connecting pipes to be leak-proof while flying, or maybe the expansion will be so big that the material will break when the engine reaches some temperature. So you try to design a solution to that problem. Maybe it would require a new material, which means almost inevitably years of research and a lot of money. Maybe you could design some very specialised connection that can operate at different temperatures and modify its geometry so that it doesn't leak at any temperature; which again may take a lot of time and money in research.
So, here comes the brilliance of the solution: You turn the design around. You define the connections' "normal state" to be the one at high temperature. You design for that, make sure that while in operation, there's no leak, the connections don't overpressure, the fit is right.
But obviously then what will happen is at low temperature the contraction of the material will produce leaks. But you think about it and realise that this only happens at landing/take-off, for the "brief" time that the plane is moving on the ground, which is both a much safer situation and a shorter amount of time. In those circumstances you can afford to have some leaks without it being so much of a problem.
So, it's not so much that the solution is "leaking fuel is brilliant", but that the design acknowledges the inevitability of a leak, and then moves the leak to the moment where its presence is less problematic. So you sacrifice some non-dangerous fuel loss on each mission, while allowing for a better, more secure solution on the moments the plane operates nominally. And all without needing those years/money spent on complex research.
1. There are no separate fuel tanks on the A-12/SR-71. The body of the aircraft is also the fuel tank. If there were internal fuel tanks separate from the body of the aircraft, then there wouldn't be any leaks. But then that would add weight and bulk that they couldn't afford.
2. The fuel they're burning is JP-7, which is so hard to ignite that it will put out a match, or if dumped on a fire, it will put out the fire. You have to get it to much higher temperatures for it to ignite. So, leaks that do happen at low temperatures (on the ground, early in the flight envelope, etc...) just aren't that big of a deal. There's not many things that would be able it ignite the fuel when it leaks at those times.
Because of this two books are interwoven in my brain, it’s a weird mashup.