A long time ago I was involved with an ‘end of first generation’ EFIS refit, which was developed almost entirely on systems rigs (sunlight brightness, flightdeck shadow tests had to be flown).
The equipment manufacturer used a fixed base generic flight simulation rig as part of their validation and evaluation process. The test pilots ‘flew’ this rig.
The aircraft manufacturer used a systems integration rig using aircraft hardware, sensors, elect system etc; it had a very limited ‘flight’ capability, but did allow autopilot coupling.
Between the two systems rigs, all of the basic development and assessment tasks were completed, to the extent that the regulatory certification team agreed that it was not necessary for them to fly the aircraft. However, they quickly withdrew that decision, most probably because the test pilot enjoyed the flying and a flight would cover any awkward questions from above. Several manufacturer evaluation flights, only one cert flight.
Later, a major digital avionics development used similar rigs, but the extent of the retrofit and projected capability (Cat 3) required extensive flight testing.
IMHO the 'rules' you seek are those which go into your certification plan. You could save money on those aspects which you feel comfortable not flight testing, but the more you choose then the greater the risk of having to fly for trouble shooting, and to improve the confidence of the certification authority which might take a big hit if your original plan fails.
Relatively recent amendments to CS 25.1302, Human Factors, might create as much concern as would technical certification under 25.1309.
I suspect that those who have done ‘it’ before might take greater risks. The design team I worked with had done a lot of ground breaking work before, but they didn’t take many risks. There was a history of learning from in-flight testing, and an acceptance that everything can throw-up something new, even with ‘off the shelf’ equipment.
The complementary view is that a rig (SIL) is an essential item as there are issues which can only be resolved / demonstrated by such a facility; you have to spend your money anyway. All you save is in the ‘cost’ of risk, i.e. without the SIL, development will cost a lot more.
The same with not planning to fly sufficiently, but here the development cost might cost more.