If you only need the one gear, put two sets of gears together and make your own shift keys from key stock. Pita to get in gear, but doubles the gear strength.

What dia pulleys are you using? May need to spin the Trans faster to take torque off the Gearset. Wot on rough stuff and clutch dumping will lead you to your current problem sooner than average.

With reducing flex in the valvetrain, you reduce "lost" cam information.  Which in turn translates into getting all that information from your cam turned into getting the valve open when the cam designer intended.  Valvetrain geometry plays a large role in reducing stresses and getting the most out of your combination as well.  Just slapping on roller rockers doesn't mean you will gain anything, in fact could be worse than what you already have if not setup properly.

After much searching, and a little experience with both, I'm unsure of which option is really better at 5500 rpm and less.  Darn sure cheaper for the dogbone, plus less rotating mass.  Yet less reciprocating mass with the arc weights.  Seems both are"wasted energy" but at least the arc weights aren't losing energy converting rotational energy into linear motion.  Just trying to figure out which one is better for power.  Dollar gain isn't much of a measure, but I'd bet most think the arc weights are the better solution. What are your thoughts?

Something that I've seen and thought worked quite well, was using the same type rubber straps used in holding a snowmobile hood closed. 

I'm not overly familiar with the oppy, but as you are intending it for commercial use, I would recommend a larger ring gap .008" or so. You will be operating in an environment that doesn't promote cooling due to the buildup of foreign matter impeding air flow.  I might suggest using the flywheel key as well, just to keep that flywheel from creeping timing.  Racers are always checking and adjusting things, not something you want on the job site.   I believe there is something to flipping the top rod over as well, maybe someone will chime in on that one.
  Everyone has their opinion on lapping valves, but I tend to the theory if things are done properly, lapping only wears more material off the valve and seat. If your angles and concentricity are correct, they will seal, and seal better once worn in.

Honestly, your best results are going to include getting your rods/pistons and crank out on the bench so you can take some measurements.  If you don't have the tools, ask around or find a machine shop if they'll do it for you.  You need to have either id snap gauges, or id bore mic to measure the bore of your rods.  Od mic to measure each journal of your crank throws, not only for od, but for out-of-round.   While you are at it might not hurt to check if your crank is bent.  Your posting seems to indicate this may be beyond your experience, so I'd suggest someone that's familiar with these tools and checks to do them for you and explain along the way.

Yes, therefore you know when you have excessive blowby!   

Well, taking what I have for measurements from the various engine components I have to choose from and punching them into desktop dyno, yup.  The AVS rod and piston combo is the way to go. Nothing new for news here.  Although quite a bit to be gained from changing the timing events of even a stock camshaft.