You have addressed all the common problems (with Bart's Z wobble fix, etc.). Here are some outliers that might be worth checking:
1. Check that your filament feed is smooth and consistent. High resistance can pull the X-carriage and head up very slightly, resulting in higher layers. When the resistance reduces, the head moves down again squishing the layer. The result looks very similar to Z-wobble. Filament pools are not balanced. Mount your filament spool on ball bearings, then add just enough rotational resistance so the filament doesn't unspool by itself. Avoid kinks or sharp corners in the filament between the spool and the machine.
2. Constrain the top of the Z-rods (
http://www.thingiverse.com/thing:119997) to stop them moving around when using the floating Z-fix with Z couplers if you haven't already done so. Not much impact on build parameters, but stops them wobbeling too much during travel and getting knocked around when working on the printer.
3. The necking-in (narrowing of part after first layers) happens when the part cools down too fast. Create a draft shield around the bed. Ensure the bed stays hot during the whole build.
4. Experiment with your Slic3r variables, including travel speed (slow it down - try 40mm/s, with your first layer at 50% speed), extruder calibration (
http://reprap.org/wiki/Triffid_Hunter's ... tion_Guide), and don't be afraid to adjust your e-steps/mm manually. In Slic3r you can also increase/decrease material volume/width for particular build sections, e.g. first layer, perimeters, fill, etc. in the Printer Settings/ Advanced tab (leave that for real fine tuning though).
5. Change belts and pulleys from MXL to GT2 to eliminate all remaining backlash - this only creates a very marginal difference though, noticeable only if your machine is already very well tuned.
6. Go back and tighten everything, again, and again. Just be careful with the Delrin wheels as the material creeps under constant load and will develop flat spots. If they are not too severe, they disappear again. If you won't be using the printer for a while, loosen the eccentrics.
7. Experiment with different filaments - just when your printer is working like a dream, you try another filament and it it will need e-steps, temps, retraction and extra-length-on-restart variables tuned.
8. A final non-scientific observation: I could swear machines need a decent run-in time before things stabilise and run consistently... give it time and keep trying.