Looks good. Your neighbor functions should be fine as long as they work for you grid(which is aligned indexed differently as you said, but same concepts).
One thing I would say about your neighbor checking function would be that they seem overly complicated for what they need to do. You know the max grid size of your map, whether it is 16x16 or whatever, so use that as M and set it at the beginning of the blueprint if you have multiple sizes. Have a variable set to be your Current index and that will be your reference. Then it should be as simple as +/- 1 and +/-M.
Also, make the base formulas into pure functions that simply outputs the index for whatever tile. Then, when you have the pure functions, make functions that are again only dealing with the index, but taking into account the wraps, edges, etc. Then that function will have the index as an output, and you can put that function in front of any Get Array node that you have. At that point, you should never have to think about your neighbor checking functions again.
Yes the world location of the tile. Just fill that vector array up and you can use it later to spawn tiles or as references for any location related action you take in regards to the grid(moving units, adding props, etc). Then you could have an array that represents whether a tile is an obstacle or not, whether a tile costs more to move through, etc. In your case, you probably want an array that says whether a visible tile is spawned there or not. You can zip through that array with a ForEachLoop and on the indexes that say yes, it leads to the spawn node which has your vector array pulling indexes from the ForEachLoop.