I'm with Cessna on this, "maximum tested to" in civil aircraft is only a limit if it's explicitly stated to be.
Creation of a justified (which probably means a bit conservative, but not too much) extrapolation off existing performance data, and working to that seems justifiable, although if you're operating in a regulations (e.g. AOC) environment, I'd suggest putting together a robust justification of the approach being used and run it past your local regulator for approval. There are plenty of standard factors available for extrapolation of aeroplane performance data that can be used for the purpose, such as some of the UK Pink AICs.
At first stab, I think that I might run the performance calcs for 15, 25 and 35°C (ISA , ISA+10, ISA+20) assuming they're on the charts, fit a quadratic to that, and then extrapolate the quadratic out to ISA+30 and ISA+40, then re-do those calculations using the standard factors in UK Pink AIC
127(P 110)/2006 as a sanity check.
https://nats-uk.ead-it.com/cms-nats/...6_P_127_en.pdf
G