There are two approaches to this; one is to provide as much raw data as possible and the other one is to synthesize it to help certain forms of understanding. They are not exclusive.
ElSupreme's list is interesting as raw data; i would add
- Ascent /descent rates 5& 10 sec for those of us near hilly areas.
- Tire pressure if available
Tommy's synthesis seems a useful one.
I would hope that just like apple/google improve maps based on devices sending in data; that Tesla can optimize range calculations by taking in data from vehicles to understand both the vehicle and road components. If I commute certain distances at certain times with certain battery states, car loads, traffic conditions, weather, driving styles, etc etc they can get readouts of actual kWh used for the trip. Would love to know that data is being anonymized, chunked up, aggregated, and digested to inform my own, and others', range calculations.
In biosurveillance we do this sort of thing when modeling complex stuff and it can be surprising how using some real-world factors to change otherwise linear models yields more realistic prediction.
OTOH, right now all I really want is all their energies focused on the production line throughput & quality!