Sounds like fun!
Here’s a planning worksheet that helps you get an equal density for the number of pixels you choose.
The number of lights to use depends on the tree height. I bought a 9’ tall artificial tree years ago before I was into addressable LEDs. It came with 900 incandescents and I replaced them with 1000 x 12 V WS2811 bullet-style. Here’s what it looks like (and I still recommend the concentric ring mapping approach mentioned):
At that density / conical surface volume, I’d recommend the 10cm spacing unless you want to buy 2X as much. I injected 12V at the 0, 370, and 700 pixel points (which was driven more by the breaks in the artificial tree segments, since I leave them attached between years). The pebbles have 22 AWG vs the 18 AWG on bullet bulbs, so you still need to inject every 200-300 pixels, which is kind of of annoying since less injection work is one of the prime motivations for using 12V!
They do run hotter than the 5V, but not uncomfortably so. It’s not the peak driving heat so much as the quiescent current that keeps them at a warmer baseline than typical 5V pixels. The pebbles you mention will be much cooler than my bullets (41mA max vs 8mA max per pixel). I don’t have quiescent measurements. I run them well below 50% max brightness for aesthetic reasons, and like all WS281X, I’m annoyed at the low-intensity chunky color transitions to off.
If redoing it today, I’d used either the pebble strand you refer to (less bright, likely less noticeable 0-to-1/255 transitions), or for the classic bulb look, they finally have 12V GS8208 bullets which have great low-end transitions. I don’t think you’d notice the black-vs-green wire in a tree to be honest, whereas I cringe at the low-intensity color inaccuracy and chunky transitions to black every year.
Using 1000 WS2811s in one serial line will result in most patterns running around 30FPS. Use an output expander from the center, or 2 PBs in sync mode to double that.