For example, there is no reliable data for stalls at high altitude, at high Mach No, at very high AoAs, or with large amounts of sideslip.
Why not? Because these operations are too dangerous even for a test flight?
No test pilot will have ever put a large civil aircraft willingly into the very high AoA experienced by AF447 - it would be far too dangerous.
Are there any projects to try pilot-less testing, at least for certain dangerous flight tests? With everything being fly-by-wire and the experience with pilot-less drones this seems reasonable.
Might get us a lot more data of the edge-of-envelope sort.....