[argyllcms] Of ink limiting and maximizing gamut

Hi,
In my experiments with Gutenprint and linearization I've come up against an interesting issue with a trade-off between ink limiting and gamut.
Since one of my goals is to maximize achievable gamut, I've so far been working on the assumption that it's desirable to maximise the amount of ink that can be deposited for each channel - thus in my experiments on 100g matte inkjet paper, I've determined the maximum input level at which I can print a single ink before the paper starts to saturate.
I've then generated and read targets exploring each channel and created correction curves for those channels - thus the printer is now linearized with respect to L* (or b, in the case of yellow ink).
The problem is that since each channel is driven to saturation, the ink limit for the paper is essentially 100%, and attempting to create a CMYK profile for the printer using Argyll with such a low ink limit gives very poor gamut.
Reducing the maximum density of each channel to make "headroom" for a higher ink limit certainly improves matters - but I'm confused as to why - since [100,100,0,0] for instance with channel densities reduced to 50% should actually print the same colour as [50,50,0,0] with the channels at full strength. Logically there should be *more* shades available at full strength but 100% ink limit than at half-strength and 200% ink limit.
Just for the hell of it, I tried wedging a naive normalization filter into my print chain (basically, divide each value by sum(C,M,Y,K)/max(C,M,Y,K) - so there's never more than 100% ink in total) and profiled through that using a 300% ink limit. The filter did horrible things to the device linearity of course, but the gamut increased significantly.
So how should I make the trade-off between gamut and ink limiting? If I need to reduce the maximum channel densities, what's the lowest "headroom" I should be thinking in terms of leaving for ink limit in order to maximise gamut?
And are there any tweaks or switches I can play with to try and improve how well Argyll can cope with low ink limits? 

Any insights or comments much appreciated :)

All the best,
--
Alastair M. Robinson

Other related posts: