I play the piano and also practice other non-musical skill-based activities. What I’ve found is similar to your experience: often taking a break (days, weeks) results in fundamental technique improvement even if we regress in terms of how fast/how accurate we used to be. These days I take advantage of this by working hard at a particular problem for a few days, letting it sit on the backburner for a few days, and then coming back to it. I find that usually even if I’ve regressed on the specifics, my technique has improved in some fundamental way such that I feel that given just a few days work I can surpass where I used to be. I work on several things at once, cycling them in this manner.

Since we’re playing the long game when it comes to any skill-based activity, I find this method more efficient and less tiring than trying to perfect one thing before moving on.