Here is the test setup I used to try and figure this out. I printed up some 13x 19 lens focus charts from http://www.graphics..../res-chart.html, laminated them and taped them to a 4' x 6' pvc frame placed in my pool.

I shot a sequence of apertures; 4, 5.6, 8, 11, 16, 22 at 17mm, 29mm, and 40mm for each of four configurations with my NX fisheye dome.
No extension ring- no diopter
Ext 20 – no diopter
Ext 40 – no diopter
Ext 40 - +2 B+W diopter
I shot at ISO 100 with AF on, targeted at the center point of the center lens chart. Exposures were adjusted manually and the testing took place in the bright Arizona sunshine. My strobes were used at fractional powers to provide some fill for the lower half of the pool. After the run the images were loaded into Lightroom and keyworded with the configuration notes. LR is a brilliant tool to use for this kind of work. I was able to easily filter for an f-stop then use the comparison screen to put two 1:1 and 2:1 crops side by side to decide which appeared to have the best focus/sharpness.
On my initial runs through the images I was primarily looking for sharp corners. Some of the choices are obvious and others required some pixel peeping to decide. All of this is pretty subjective on my part. You might make a different choice in some cases. The images I've included are 1:1 crops of the test chart in the bottom left corner. Might be a 2:1 in here somewhere. Here is what I think I found;
-The center is sharp in all cases I tested
-f/8 and f/11 is the sweet spot with markedly better corners than even f/5.6

f/4 is not worth shooting if you can help it.

f/8 to f/11 is a touch better than f/16-f/22, which I didn't expect. 2:1 crop

The Sea & Sea recommended configuration provided the worst corners in all cases. The 40 Ext on left, no ext on right.

To be continued:






