This is a great summary!
I can only confirm your findings. Those are very similar to my own conclusions. For example I very much prefer the different files for different passes as well. Pausing and changing toolheads would be more complex.
I too take pictures of the screen with where the original work-origin was set as I’ve made that same mistake 
of accidentally resetting it when I only wanted to change the z-axis.
Some personal additions:
- For setting the z-height: I use a small piece of conductive tape (copper or aluminium tape, the last one you can get pretty cheap, it’s often used to seal HVAC tubes) that I paste on the spot of the workpiece I use for the z-level. Typically close to the border so it hangs off on one side I i can connect a small alligator clamp to it connected to my mulitmeter. The other one connected to the drill bit. Set it to beep on zero resistance and lower slowly until it beeps.
- When you do a run boundary: always go up a few cm before you do so. First time I usually also make sure that the drill bit is higher than the highest clamp. Just to be sure. Run a second time lower when you’re confident it won’t hit anything.
- After setting the z-axis height for the work origin and starting the job: always raise the toolhead again high enough that it won’t hit anything. Otherwise it will just move horizontally and a small uneven-ness in the surface of your workpiece could cause a scratch, your bit breaking or the linear module skipping steps
- If you won’t need to reposition your same workpiece on the table, a really simple way to mount it is as follows:
- put some strips of painters tape on your wasteboard.
- put some strips of painters tape on the opposite side of the workpiece (aligned with those on the workpiece
- use some superglue on the painterstape (not too much so it doesn’t go over the sides of the tape)
- stick them together and keep pressure on for 20 seconds.
- by the time you’ve set your work-origin you’re ready to mill
- “tram” your bed and linear modules in a repeatable way.
- You can do it properly like @brent113 does here: Carriage Tolerances - Unusable Over Distance >75mm From Center - #34 by brent113
- or the way I do it. Before starting a milling project: when the power of the machine is off: raise the x-module completely to the top (pull/push it up as far as it goes on both z-axis) & pull the bed completely to the front as far as it goes. That way you know both are aligned (not needed for the x-axis as there is only one). Maybe they are not aligned perfectly, but at least it’s consistent and repeatable. If (like me) you’re a bit too aggressive with the speeds and feeds in your first projects get something stuck. You can make sure this way that you can get it back to the same alignment as before. (just repeat the above procedure)
- Make sure your bed and x-axis are aligned & parallel to each other. It’s not necessarily a big problem for some projects and if you’re cutting the top layer away anyway. Bu I’ve done a 2-sided project recently and in that case it’s really important. The easiest way to achieve this is to mill away the top 1mm of your waste board.
- Always make sure you can turn the machine off immediately (kill power!) Again, for the above situations. It will happen at a certain point in time, especially if you’re new to cnc (like me)
- Use some form of dust collection. I’m using this setup and it works really well: CNC vacuum (too much wood dust) - #8 by brvdboss I did put in a (3d printed) cyclone dust collector in between and that works really well. (using it with a standard home vacuum cleaner at the lowest setting and it works really well for me)
- When using a setup like this: make sure you don’t run the attachment in your workpiece or clamps. I’ve made the mistake of doing a run boundary without it being mounted and then hitting clamps when actually running the job.
- For the toolpaths:
- So far I’ve preferred speed over depth of cut. Depends on the material obviously (in my case mdf mostly so far). When in doubt, I prefer a less deep cut and more passes at a potentially higher speed. The advantage is:
- smaller risk of your bit getting stuck in the material and breaking
- If you want to go faster at some point, you can just increase the speed on the touch screen. Going slower with a deeper cut doesn’t work that well usually, and you could end up “burning” the wood in stead of cutting it.
- A nice side effect of bumping up the speed: the travel is between different spots goes faster as well. With the free version of Fusion the travel speed is rather slow (one of the limitations of the free version)
- Some materials are harder/softer at different depths (again mdf: hard first mm, softer on the inside). That’s when I usually increase the speed a bit after those first layers.
- When starting: start slow the first time with a new material/bit and adjust speed based on what you hear
- When I was starting I often set the work-origin 3mm higher and watch it do a first layer before I started at the real hight. Just to get a feeling for it.
- So far I’ve preferred speed over depth of cut. Depends on the material obviously (in my case mdf mostly so far). When in doubt, I prefer a less deep cut and more passes at a potentially higher speed. The advantage is:
- I’ve tried some pcb milling as well (tiny 0.2-0.6mm bits). If you want to avoid breaking too many bits: get a sheet of xps-insulation foam and experiment on that. It’s really easy to machine and getting speeds or depths wrong won’t kill your tools. Do make sure to have a vacuum, those things fly everywhere.
- For two sided milling: I use the “dowel” method. Create a toolpath with 4 holes around your workpiece. I’ve cut 20mm segments of a steel rod (diameter 6mm). Machine holes the same size in the wasteboard (or a sacrificial plate in between). Machine holes in your workpiece (outside the border of your actual resulting piece obviously). Depending on the thickness of your stock you can do that in one go or to do it in two passes. Put the metal dowels in. Machine one side. Turn your piece around, put it back on your wasteboard using the dowels for alignment and machine the second side.
- In all honesty, I didn’t get perfect results on that yet. There was some misalignment still, but it looked like there was some torsion on my vertical axis. I still need to do some more testing with this.
- The final cutout (which I did do last in this case) I thus did from one side to avoid that misalignment. (that way you can’t really see it with the naked eye. it’s really limited, but still it avoids the need of sanding it away afterwards)