1. This should be adaptable to the SD4 with little trouble - everything is approximately the same size as the SD3.
Ball bearings are good - over time the original bushings wear, and the friction increases. You may not see any difference in new machines, but the ball bearings last much better, and only need very light lubrication - this prevents that abrasive mix of grease and other stuff forming...
New rods is a good idea for the ball bearings, as there has been some consternation about the exact size and surface hardness of the original rods - the ball bearings are much more sensitive to this! Although it costs more, you'll have the best experience and get the most mileage if you replace the rods and bearings together.
I'm not familiar with the SD4 y-axis belts, but I hear they are continuous and clamped on, and tensioning is done from the front idler blocks. This is potentially a better solution than the punched-through-bolt on the SD2/3. 2.03mm pitch x 6mm wide MXL belts is available widely by the m/ft on eBay and 3d printing websites for basically nothing (a couple of $ / m at most, and you need less than 2m to do the y axis), so you could easily purchase a length of new belt to try out the lawsy carriages without cutting the originals.
2. Haven't done fishing line myself, but it's well regarded - talk to wardjr. 
Pretty much no different to doing the above, but use line instead of new belts. Only addition is you have to measure and update the steps-per-mm.
3. I've never come up with a good solution for this. Please experiment, and if you're happy with it, post it up here to show us! A new computer fan may not cost much, but note that you may want to connect the nozzle fan for g-code control - so if the extruder fan is keeping the barrel cool (which can help to reduce clogging), then you might want to keep that there...
4. It's pretty cool. It won't improve overhangs really, that's all about cooling and part design. If you're happy, no need to go to the trouble of changing (it's wiring and a firmware update). If you have trouble with your hot-end in the future though (e.g. continuous clogs, catastrophic melting failure, etc.) I'd highly recommend it as a replacement though.
5. I avoided the threadless ballscrew as I didn't really like the assymmetry of it (it's steps-per-mm actually will vary with direction and to a small extent load - this makes me uncomfortable!). Replacing the z axis is a bit of a big job, you end up with most of the printer's guts on the floor. Consider the M5 / M3 threaded rod also - this also works really well, and is a bit easier to get to grips with. Also important for the z-axis is making sure you've dealt with the backlash: I had quite a lot, and the first few layers were always a mess. Using the M99 'hysteresis fix' to correct in software worked wonders for me.
SD3. Mk2b + glass, heated enclosure, GT2 belts, direct drive y shaft, linear bearings, bowden-feed E3D v5 w/ 0.9° stepper
Smoothieboard via Octoprint on RPi
