1) Is that to simulate inaccuracy? Apologies if I'm overstepping, I find the simplest way of handling target prediction is,
aimPosition[x,y] = enemyPosition[x,y] + (enemyVelocity[x,y] - yourVelocity[x,y]) * (distance / weaponSpeed)
Which, ignoring acceleration, provides the perfect firing solution - then I add an inaccuracy factor on afterwards.
3) Yeah, stronger rather than more enemies will help, although it'll still likely lag due to the vast number of turrets and projectiles flying around. You could use a fixed time-step; the game would slow down as FPS drops, but it'd avoid the glitching through walls. Pick your poison.
4) I've been in that exact position, needing that functionality, but can't work out how to use the junction. I stuck it in the middle of two conveyor belts in a cross formation, but the resources just stopped at the junction and started to pile up. Maybe I'm just being dumb, but it could do with an example. Perhaps bung one in to the initial base setup at the start of the game?
5/6) Yeah, totally understand. Though I personally find the online highscore the most complicated bit to implement. I don't know what tools you're using; I usually put something together with socket.io, node and a small AWS server - but the score verification is always the hardest bit - it's really easy to spoof a highscore in some games.
+7) I also forgot to mention, not a bug necessarily, but when you're building things and run out of a certain resource, it auto de-selects that object, which is a little frustrating cos you have to go back and re-select it each time. Also your gun starts firing on that final object placement as you run out of materials.
Jeez, sorry if I'm bombarding you. I'm a sucker for resource management / tower defence games. Do you mind divulging what code / libs you're using for this? I've been mainly using phaser.io recently but I'm not sure it'd handle so many sprites as your game requires.