Highlight extension is different to HL DR... HL DR does exist, it's valued by the exposure required for it to clip up to an unrecoverable level... A JPG file, if it is the outcome of an unprocessed Raw, is a good hint for the behaviour of a sensor if HLs clip on one sensor while hold up on another for the same exposure... "Linearity" of modern sensors is also questionable... None knows how makers "filter" (in the A/D conversion) the output voltage from the sensor transistors... we presume it's linear, but ...is it? 
There is for sure a limit to the amount of illumination a sensor can absorb before saturating (is it called well saturation?), but:
- there is absolutely no way to measure this looking at the 255 discrete values cooked into a jpg file. The only aspect you can assess is the exposure relative to this threshold, meaning that you can at best get some visibility about the way the ISO value was set relative to the saturation threshold mixed with the curve that was applied by the jpg conversion engine,
- linearity has been proven time and again,
- none of that determines the range of illuminations between which the sensor is able to capture useful imaging information, which is called the dynamic range.
So, would it only be because the R in DR is range, highlight DR does, by definition, not exist.
But who am I to object if your equipment purchase strategies are set based on jpg analysis of highlight DR.
Cheers,
Bernard
