Suggestion: allow us to choose focus of "Strengthen Skills" button.
There are two main ways I've seen people use "Strengthen Skills": as generic practice and as an attempt to keep their trees golden. These approaches require a different mix of questions for the exercise.
If you want to just practice what you have learned, you want a mix of questions from all the skills. This will bring up more content; however, it will not help you increase the level of your skills much. That's because you'll only be touching one or two words from each skill and that won't be enough to flag the entire skill as having progressed to 'gold' status. This seems to be how the old setup of "Strengthen Skills" worked.
On the other hand, if you want to keep your tree golden, you want each session to focus as much as possible on one or two skills that are at less than 5 stars rating. You want to see progress when you complete the skill. The trade off is that each session will have a very narrow scope and you might not see material from parts of your tree in a long time.
What do you guys think?
If you want to strengthen just one skill at a time you can go into the individual lesson to do so.
The spaced repetition algorithm will take care of your last concern so there really is no need to do the generic revision of the whole tree unless you are happier with tha. My personal preference is to revise individual skills when prompted by their loss of gilding and I rely on the spaced repetition to remind me when to practice those skills.
And that's the thing - the algorithm for question selection on a generic revision of the whole tree today is targeted specifically at the second scenario. There were a lot of complaints a few months ago that people would do a tree revision and not see any skill improve so the tree revision prompt was tweaked to focus on one (or two) skills so that you could see progress.
This tends to work OK if you only have a few skills that are not gilded. However, if you don't keep up with regilding your tree, the algorithm's preference for weighing more heavily on early skills starts to be noticed. I personally don't have a problem with this, but I have seen other people complain about it.
It's particularly strong if you stop working on a tree for more than a month. When you come back and all your skills are at either level 1 or 2 the algorithm will push strongly push a top down approach to reviewing the tree.
The thing I sometimes dislike about strengthening individual skills is that it's kind of predictable what sort of grammar you will be tested on (e.g. conditionals in the "Conditional" skill). Having a bunch of different things thrown at you keeps you on your toes a bit more - and you can always strengthen a specific skill if you want that to be gold.