Raising compression always gives gains, but it has a cost associated with it: you will be forced to run higher octane fuel. Milling the deck is something you do to get the squish right if it isn't right from the factory. Most of the time it is not, but modern machining is making things better. The factory wants to err on the side of too tall because the engine will run too tall, but will not run too short. Decks are usually spec'd at some number + a tolerance. There is no - in the tolerance. If you get a block that is decked correctly from the factory, you are one very lucky mug. I suspect zoltiz has one of these engines because his fuel economy is so good. Deck height and squish are called artificial octane because when you get them right, the engine is far less sensitive to fuel quality.
Milling the head is a common way of boosting compression, but the 2AZ uses angled squish pads, so milling the head could be very problematic with the stock pistons. Most manufacturers leave at least 0.010" on the head to allow a light resurfacing in case of a headgasket failure, so it's certainly possible there is something there to work with. Still, I'd want to be sure I've done all the homework to make sure I have the exact clearance I want (somewhere between 0.5 and 1.0mm squish height). Too tight means the pistons will hit the head. Also, because this is an aluminum block, you need to take thermal resizing into account and set up your clearances to be correct at operating temperature. Expect the block to grow significantly from cold to full temperature.
Removing the balance shafts and lightening the crank would be excellent for improving power delivery. Neither of them will "create" horsepower, but both will liberate power currently diverted to some other function. It would not suprise me if there is 10 pounds of excess mass in the 2AZ crank. Removing it and the balancers would do wonders for throttle response. The big HOWEVER is removing the balancers will make the engine a lot more buzzy. Big fours are inherently not well balanced, and even a full precision balancing will not eliminate the tendency for the engine to shake. Expect more vibration in the car, and shorter engine mount life if you choose this path. I wouldn't do it to my daily driver, but it would be one of the first things I'd do to a racer.
I don't need to say any more about a lightened flywheel. I have one installed. It's a major improvement IMHO.
Anything that takes the engine's mechanical spec away from what Toyco intended will result in changed tuning. Reflashing the ECM or piggybacks will be necessary for anything other than blueprinting to OEM specs. So far, I haven't seen anyone successfully reflashing their ECM, although Scionspeed made veiled claims about this a while back. They don't seem to be doing very well from a business perspective these days; there are a lot of pissed off preorder turbo kit buyers out there who would say they really suck, but I have no personal experience to say one way or the other. Their parent company, Jamasco, has done some very interesting stuff, and they appear to be quite technically competent, but their business skills seem to be coming up short these days.
It would be great to have a junkyard engine to build. I can think of lots of things I'd like to check out, especially the port configuration in the head to see how it would respond to port work and ceramic coatings. That will have to wait until I get a head sitting on the bench in front of me.