I agree with Denny that it should not matter much in reality, but in theory at least I'd say that the sparse coil may be the favored option.
The effectiveness of your chiller depends (amongst other factors) on the steepness of the heat gradient or, if you will, the temperature contrast between the chiller and the immediately surrounding liquid. The larger the temperature contrast the more energy can be transferred from the wort to the cooling water per time unit.
The wort will be chilled radially from your tubes and if the coils are closely packed they will together cool the immediately surrounding wort rather quickly, thus decreasing the heat exchange. A sparser coil formation will reduce this cooperative chilling of only the wort close to the tubing and increase the heat dispersion throughout the entire liquid.
This effect is the entire reason that using the whirlpool feature (or just plain old stirring) is so beneficial and the stirring of the liquid will have a much greater effect than the spacing of the coils. Even so, when I built my own i made sure to space the coils as much as possible while still making sure that the entire chiller would be immersed.
In conclusion: I vote for option 3 for its combination of simplicity and theoretical function.
Edit:
The coil itself does not center any trub, its the movement of the liquid.
The velocity of the moving liquid is much greater at the edges of your pot then in the middle and so the heavy particles gathers in the middle as it would require more energy to accelerate them than the surrounding water.
The cooler would thus be more of an obstacle to trub formation than a help as one might first believe and so yet again the sparse formation would be the better option. However with this in mind your legs in figure 6 might not be a bad idea.
Once again however: I don't think that all this thinking will much improve the final product or greatly simplify any part of the process, fun as it is.