That said sub regions could work - but I think they need to be part of something bigger, so for example using Granada as an example in the North West if they also had Granada Manchester and Granada Liverpool those stations IMO would have a much greater chance of succeeding than independent Manchester and Liverpool stations.
Isn't this where the UK model diverges from the US model?
In the US the geography of TV transmitters (plus the popularity of cable) means that there are often transmitters (and thus local stations can exist) per town - as the towns are far more spread out - so the hyper-local model works. In many parts of the US each town/city has a dedicated TV transmitter, or transmitters - and local stations as a result.
In the UK (and Europe in general) we're so much smaller as a country that single transmitters cover lots of towns and cities (as the town and cities are far less spread out), and our terrestrial network is based around providing universal coverage of networked stations (and we achieve significantly higher percentage population coverage on terrestrial as a result)
As a result - editorial or town-based sub-regions don't map neatly in the UK (and to a wide degree Europe). To quote your example - Granada Manchester and Granada Liverpool sub-regions aren't possible with a single transmitter (Winter Hill) covering both areas - unless you were to allocate separate muxes to each region (which is expensive as it is duplicating spectrum)
If the TV landscape were more cable-centric, then you could engineer more localised services, but we're not.
Europe has historically gone down a national network with regional variations model - and I, personally, don't see that changing. If anything I see terrestrial TV becoming less and less varied - as the RF spectrum is squeezed, with local services being IP-based. (Look already at the move to put IP stuff on Freeview HD) Making money out of local services is the other issue I guess...