1. S&R dispersion measurements are normalised to on axis frequency response, not on LW. Since Type 20 is made to compensate for diffraction, of course it looks worse.
The normalized spectrogram gives information about how good the directivity of a loudspeaker is. This directivity can no longer be changed by any "normal" measure. To change it, the crossover or the cabinet would have to be changed.
Therefore, the normalized spectrogram gives information about how much effort has been put into the design, selection of the drivers, design of the waveguide/horn (if available) and the tuning of a loudspeaker.
Here is an example of a loudspeaker with bad horizontal directivity (I'm ashamed of it) and how this bad directivity has to be compensated during crossover and tuning of the speaker to not sound like sh..t.
If you do everything right and, for example, compensate the strong expansion of the radiation in the range of 1.5-4kHz by lowering the on-axis FR (create an wide dip between 1.5-4kHz), you will get a loudspeaker that, all in all (non-normalized spectrogram), radiates quite balanced.
So if you write Hepp Type 20 "compensate for diffraction" this is just a desperate attempt to compensate for a non-optimal radiation by adjusting the on-axis FR.
Don't get me wrong, this can still sound great in a normal listening room, but as a working tool in an audio studio you would want something more neutral.
Compared to the KH310 I would expect that the LW is more balanced there than with the Hepp Type 20, because for example the vertical radiation of the KH310 should be more linear at the transition from midrange to tweeter.
Maybe Amir will get both speakers to measure, then we will have certainty.
2. Measurements in S&R are done without Lineariser.
As written above, it is not possible to correct the errors in the radiation, they can only be compensated by changing the on-axis FR.
When the "Lineariser" linearizes the frequency response on axis, it gets exactly the radiation that the normalized spectrogram shows.