The LT3080 and LM317 are voltage regulators meant to power a fixed circuit. They have a limited accuracy because the power dissipation heats up the circuit and reference. In addition they have limitations on the load impedance and may oscillate with the wrong output capacitance / load, which can be annoying for something like a lab supply where you don't know upfront what is connected.
Well, the datasheet for lt3080 has performance curves. Looks like reference current is is ~600ppm/c, which is about 2.5% for 50C raise. I'd say it's fine for an analog power supply.
As of oscillation and stability, what makes lt3080 special? Just put 10uF ceramic cap to the output. LT3080 is a very fast regulator, so I can imaging stability issues may happen (got that with a DIY load), but that's the price to pay for speed.
LT3080 is expensive, but if you take into account time and effort engineering a cheaper circuit it's actually not that bad. Unless taking an existing design. Also, I bet there will be much more parts.
PS this is my LT3080 power supply
https://www.eevblog.com/forum/beginners/please-critique-this-linear-psu/ , and this is a picture of it:
https://www.eevblog.com/forum/testgear/show-the-homemade-equipment-you-are-using-now/msg1984904/#msg1984904 . It has a horrible temperature drift thanks to voltage set circuitry. The problem is not with lt3080, but with lm334 which has a horrible drift. It is so bad that the datasheet proposes to use lm334 as a temperature sensor...