You are basically correct that the dual rotor machine does have a closed magnetic circuit. The air gap is much wider than with a slotted or toothed core where the windings are wound in slots, so typically you will get about half the flux density in the gap compared with a slotted core. This type of machine has only become possible with the introduction of neo and samarium cobalt magnets.
Now to the issues that joestue raises. He is referring to rather specialist machines which are somewhere between a conventional slotted core and the true air gap ironless machine. If you are new to this then they may be well outside your range for a first attempt. The toroidal slotless machines are a good compromise between the conventional core and the air gap machine and if used with good quality low loss strip wound toroids they should perform well.Being slotless the air gap is still large compared with the conventional slotted core which keeps the iron well below saturation and the core loss is not a big issue.They have been used for years by some wind turbine manufacturers.
I have used slotless radial designs which are a single sided version of the same thing.
If you can get the required high quality core material and can solve the winding and stator support problems associated with these then it may be a way to go, but I wouldn't consider it to be a beginners project unless you have a lot of experience constructing similar things.
As for iron loss not being an issue then I would accept that slotless iron cored machines well constructed will have an iron loss low enough not to be worth worrying about. This is not true of conventional slotted cores when worked at normal flux density, the loss from them is not negligible.If you live in a windy area then it may be of no consequence but if you want to extract power from 7mph winds then unless you use excellent core material and wide air gaps and relatively low flux then your iron loss will defeat you.
Poor starting of slotted iron cored machines in low winds is as much a problem of iron loss as cogging. Careful design can eliminate cog or reduce it to a point where it doesn't affect start up but it still will not pull out of stall until you develop more power than the iron loss.
If you are in a good wind area then any form of iron cored machine will do the job. If you want to use a slotted core then radial is the better choice and is normally the only option as you have plenty of motor cores to choose from.
If you can produce your own core then the outer magnet rotor radial ( drum type) is a better arrangement for wind as there is more room for magnets and no danger of them flying off.
The drum type slotless machine is a good one and is better cooled than the dual rotor, it has low enough iron loss not to worry you but if you mess up the design it is a rewind, you can't play with the air gap.
For axials the slotless single rotor is possible but core material is not normally available cheap, with poor core material it is very poor. The axial construction virtually forces you yo use a winding that has spaces in it unwound so copper use is not as good.
Slotted axials are very difficult and seem too much effort for the results you get and again core material is a big issue.
Toroidals are specialist things capable of excellent results if you can get good core material.
If you choose anything other than a motor conversion ( radial slotted) or a dual rotor you will be more or less on your own so you have to decide whether your design and construction ability is up to this as a first attempt.
Nothing new under the sun, all done before the turn of the last century. Materials and manufacturing issues have dictated the trend over the last 100 years, but materials change so it is worth keeping a n eye on the old ideas, but for a first attempt it may be wise to stick to something conventional so that you can get help from others.
Flux