I see, this is very different question then you initially posted.

Ironically the subject was discussed years back what I was fighting with S2’s “fuzziness” at very top.

I still do not feel that there are any 'rule of thumb' in it, there are rather many “depends”.  If your MF driver in your particular horn has problematic top end and you would like a bit roll it of then there are many options to consider BEFORE bringing a new amplification channel for HF. Here are a few options:

1.       Use a deeper horn
2.       Use a horn that has higher “constant directivity” characteristic
3.       Introduce a very mild serial inductance to MF driver. (HF will need own amplification)
4.       Mild application of damper on the surface of the MF defrag  (like aquapluss)
5.       Changing reflective characteristics of back chamber on MF driver
6.       Lower the output stage that drives MF loading (the more effective  and proper in my view)
7.       Use HF channel in a way the it will mask out the upper MF driver
8.       Reduce effectiveness of phase plug

KS wrote:

Thats the kind of stuff I was hoping for. I appreciate the help with this, there seems to be a scarcity of horn related discussion on the web and its proving hard to learn the ropes. One other unrelated question that could save me a lot of trial and error - If time and phase alignment was not an issue (imagine you were forced at gunpoint to use a digital XO that could compensate for this), would it be advantageous to align the mouths of all horns rather than the drivers? I was looking a building something like a Macondo frame, but I'm wondering the price one pays as far as HF from the tweeter horn smacking into the back of the horns above and bellow? I see you use a fairy directive horn to try squeeze the HF through the narrow gap, but I'll likely have a round tractrix back there so was wondering if I should move the tweeter forward and time align digitally, having that option on hand.