Need and money making rarely go hand in hand. If the 135 f2 had been a big seller there is every reason to think there would have been iterations just as there were with the superteles, the 2.8 zooms, the 85 f1.2, the 35 f1.4, the TS-E's, etc etc.
Zooms are an entirely different animal than primes. Fifteen years ago if anyone had suggested that even a higher end zoom could compete with a mid-grade prime on IQ they would have been laughed off this forum. That's no longer the case.
Super Telephotos have improved much more in terms of weight and balance, and IS than in terms of IQ.
The TS-E lenses were updated to incorporate the new variable on-the-fly adjustment of the tilt and shift axes relative to one another. Canon calls it TS Rotation.
The EF 85mm f/1.2 was updated with a faster AF motor, at least relative to the original model that was excruciatingly slow.
The big difference with most other non-ST primes has been to move from labor intensive hand assembled designs to machine assembled designs.
That, and the whole idea that has taken root over the last decade or so that the only measure of a lens' worth is how well it images a flat test chart.
The EF 135mm f/2 L already had outstanding IQ in the areas that a fast 135mm lens is typically put to use. Two-dimensional reproduction work is usually done with shorter focal lengths, or with larger formats and scanning backs.
There's no pressing demand to reduce the size/weight of the EF 135/2 when compared to more recent offerings from Sigma
et al that dwarf the Canon lens.
As telephotos go, it's a fairly simple design that wouldn't benefit as much from automation as other, more complex designs, such as a retrofocus 35mm f/1.4, do.
Compare the block diagram of the EF 135mm f/2 L
to the block diagram of the EF 35mm f/1.4 L II