Hi Ego -
First, you've got to take the OOG warning feature of PS with a major grain of salt. As I recall, it was one of Adobe's very first (ie, oldest and crudest) attempts to provide photographers with a clue that they may have OOG problems. Personally, I use this tool only to quickly point out where to look for problems, not how bad they actually are. For example, the OOG warning tool can be terribly misleading in that it shows that some huge area of the image is OOG, but when you look at the problem in more detail by repeatedly hitting cntrl-Y to toggle "Proof Colors" on and off, you see that although the area might be big, the amount by which it's OOG is miniscule. The latter is the way you should evaluate possible OOG problems.
For example, if you look at my tweaked version of your "clubhouse with turbulent sky" shot, the default OOG warning area is huge -- much of the green foreground. However, if you turn off that warning, and instead, simply toggle proof colors, even with just the default soft proofing setup ("working CMYK"), you'll see that the change in the greens because of being OOG is just about completely negligible. Similarly, if you look at my tweaked version of your silhouette shot, you'll see that the areas of concern are very small (only the most deeply saturated regions of the clouds). In this case the amount of change is more substantial, but since it's only in a small area, and, for all the viewer knows, this could have been the way the clouds actually looked. I wouldn't worry about my versions being OOG, especially if you use "perceptual" rendering.
Second, you have to realize that most short-run book printers are not using offset presses, but rather, fast inkjets that have more than 4 inks, giving them a vastly larger gamut than traditional 4 color (CMYK) offset presses. So, unless you obtained and set up a custom ICC profile from your printer, but, in fact, are only relying on PS's default "working CMYK" soft-proofing setup, you are likely being much too conservative. Some of the modern inkjets actually have gamuts larger than sRGB -- more like Adobe RGB.
BTW, here's the URL of an article on soft proofing that I rather like:
http://www.cambridgeincolour.com/tutorials/soft-proofing.htm
Third, if, for some reason, you really don't like how the perceptual rendering intent deals with the OOG areas, the absolute worst thing you can do is to dial back the saturation and contrast of the entire image. That will make everything go "blah". Instead, if you really insist on handling these areas yourself (instead of relying on the mapping done by perceptual rendering in a "convert-to-profile" step), make a feathered selection of the OOG areas and only dial them back a bit, not reduce the saturation and contrast of the entire image. Personally, I probably have had to resort to this on only a tiny fraction of my product photos, and then, the solution was usually to add a spot color ink for printing these areas, rather than attempting to find some global fix.
For my tweaked versions of your pix, I strongly suspect you would be perfectly happy if you do nothing more than get an ICC profile from your intended printer, and then do an "edit / convert to profile (advanced)" using "perceptual" rendering intent and black point compensation. Obviously, you can't do this from the GIFs I've posted (because they only have 256 colors + noise dither, and were down-rez'ed), but I'll be happy to put them in Dropbox and send you the link.
With respect to your email, I'm sorry, but I've got a bunch of paying projects going on and just don't have the time to do this for you. You obviously have a decent monitor, a good pair of eyes and a basic knowledge of what to look for, so I think you'll do it just fine yourself.
HTH,
Tom M
PS - Obviously, producing a book is a major undertaking in terms of both time and money, so, as you pointed out, you don't want to run into unforeseen problems. One very good approach to this is to ask your intended printer if he will run off a couple of test pages for you to test for gamut and other potential problems.
