Either way, once or twice a month for a few minutes seems unacceptably meagre. If it is just physical dexterity which is elusive then it might be a matter of taking the time, immediately after a successful session, to carefully analyse exactly why things went so well. It has always struck me as odd that players spend countless hours brooding over perceived shortcomings. Then when an unimaginable positive occurs they accept it with so much relief and gratitude that the rational faculty disappears altogether.
Perhaps in this improvisation you have posted you are doing patterns which fit your hand really easily and thus increases the % control you have over the procedure. When you try to do it with composed material or other scale runs of a different shape there are subtle differences which impact upon your control. If you slow down your recording you can hear the inequality of your scale runs so it is not necessarily as controlled and crisp as you suspect unless that's the effect you aimed for.
It is annoying, because I can physically feel the difference in technique. I believe I do discover efficient hand movements, but they don't "stick" well enough for me to perform consistently! I end up wondering if this whole thing is tangential to learning, and the actual learning of technique takes place through exercises and practice over a period of months or years (which I've never really done, consciously at least).
Why not try working on technique consciously for a few months and seeing what happens? I forget whether you have a teacher; if not, Josh Wright and Graham Fitch both have excellent, free videos on youtube about scale technique. Half an hour a day working on technique for a couple of months might give you some useful information