Increased PAN can be a double edged sword, too, can't it?
Apart from the issue of unwieldy class sizes, has it not be shown that class size is an issue in academically varied classes (like yer average comprehensive), i.e. academic DCs learn just as well in a class of 30 as they do 20, but the big difference happens with far less academically able DCs who hugely benefit from small class sizes (our most academic Public Schools have quite large class sizes because all the DCs have been selected on their ability to 'keep up', and do).
Another consideration:
At what point will a 'good' school cease to be if it widens its net? In the selective sector, it stands to reason that a school that admits the top 10% of DCs by examination is unlikely to turn out such high results if suddenly it admits the top 15%. Not dissimilarly, in the case of a 'good' comprehensive: in the absence of lotteries, the school 'selects by house price'. Its intake will be the DCs of parents committed enough or wealthy enough to buy in catchment. Though the effect may not be as strong as in a grammar area, there may be knock-on effects of bringing in DCs from a wider area if the wider area -let's say- includes an area of huge economic and social deprivation.
And I'd also agree that few high performing comps are going to leap at the FSM carrot as their 'fear' will be that, in the absence of interview, that could saddle them with the one or 2 pupils necessary to make learning impossible for the rest of the class.
SO not as radical at it seems, perhaps!