Latimer Alder, you say:
Which is ‘right’? In each case the conclusions depend on the starting point
The only example you provided which was clearly cherry picking was Girma’s. I believe the first example shows what is causing your confusion. It doesn’t hide any data. Instead, it’s wrong because of the logic it uses. It assumes there was a single linear trend since 988BC, and that’s wrong. It’s bad logic; it’s not hiding data.
If you want to try to figure out if something is cherry picking, here’s a simple process. First, you must be looking at a subset of what’s being analyzed, not the entirety (if everything is there, nothing is being hidden). Second, if you pick a different subset, you must get a notably different answer. That’s all there is to it.