Improving the LM3x7 regulator circuit

Linear Tech showed this circuit in the LT1085 datasheet -- easily applied to the LM317. The fixed voltage versions of LT1963A (which I prefer to the LM317) has a "sense" pin which gives you something approaching Kelvin connection.

Couple of questions for the LM317: when it is said to put the sense Resistor to the case, is it OK to just put in on the Output pin, or is the case absolutely necessary here?

Secondly, the sense line here is another connection to the load Ground, which is also connected to the Regulated power output Ground, right?
 
mAnd yet other questions, the reason being I am experimenting with a Linear PSU for my DAC and used an LM317 from my parts box. So far, from the bass to the mid-range it sound great. I do think more work is needed in my build for attack transient rapidity, and the highest ranges, so I've been reading on how to improve it.

Can the use of the transistor at the Adjust Pin (Fred) also be used with jbau's optimised values for the LM317?

Was the overall conclusion by jbau on the LM317 for attaining as flat or smooth impedance deemed a lost cause in the end because he seemed to have abandoned it?

I am thinking I could be better off using newer regulators, what do you think?

I can still further my understanding and learning with the LM317 though.
 
Hi, thanks for this very interesting thread, I'd like to ask about importance of the quality of the adj. capacitor. I mean the capacitor used to bypass the ADJ pin to GND. Do you think that it's worth to replace the adj. electrolitic 22uF to the 10uF MKT?
Leakage current would be lower and stability higher but does it matter in this place?