Off the top of my head, I will guess that XG does a fine job with this position. I have not set up the bot, and systematically checked every roll, but if I did, I expect it would play correctly.
This is a mutual holding position in a close race. Volunteering a shot is generally wrong when a constructive alternative exists. Barring a safe roll that clears one of the rear points, both sides should consolidate their positions over the next couple of turns. I don't think that strategy would be any different if you changed White's board. You might give White his 4pt, for instance, by moving a checker back from the 3pt. Blue should still play the same way.
I have heard it said that spreading blots around in the home board can backfire, because an opponent might make the mistake of volunteering a shot. If he runs off his anchor, for instance, you may feel obliged to hit, even though your board is not made up. I understand the concern, but I don't buy the logic. If a blotty play gives me the highest possible equity, that is the play I should make. If my opponent is supposed to play safe, in spite of my blots, but chooses to volunteer a shot instead, then my equity goes up, not down.
When an opponent volunteers, of course, you are not required to hit. When hitting is in your favor, you should do it. Otherwise, don't. Since I will only hit when it is to my benefit, an opponent who needlessly volunteers a blot adds to my equity. Sometimes, my opponent will get away with it. Sometimes I will lose the game. In the long run, however, I am better off when my opponent chooses plays that are suboptimal.
In your question (1), you are suggesting that XG is one of those opponents that will needlessly volunteer. My experience is that XG errs less in this regard than GnuBg. As stated above, my educated guess is that XG's handling of this position will be fine.
As for question (2), it is probably true that bots tend to spread out the blots more than human players. Even against human opponent's, however, I think you will see this sort of position crop up.
I spent a few minutes constructing retro sequences for the original position. I never found one that was perfectly played. I did, however, find a couple that each had only one, small error. The errors, however, were obvious. No human would make them.
In the following sequence, the error comes on White's final move. He should make his 4pt. Instead, I need him to "get blotty," in order to create Svilo's original position. XGR++ has the error pegged at 0.0048, less than 5 millipoints.
                         |
| ||||
| XGID=--B--AD-C-b-cC---dBe--a---:0:0:-1:42:0:0:0:0:10 | |||||
| White to play 42 | |||||
| Turn | White | Blue |
|---|---|---|
| 1 | 51S = split = 24/23, 13/8 | 64P = point = 8/2*, 6/2 |
| 2 | 64R = run = bar/15 | 51S = split = 24/23, 13/8 |
| 3 | 63R = run = 24/15 | 65@ = anchor = 24/18, 23/18 |
| 4 | 64O = outer = 13/3 | 62$ = slot = 13/5 |
| 5 | 42I = 6/4, 6/2 42P is better by 0.005 | 63 = ? |
Mike