^^^ What he said
FYI, some might do this type of mod and never think twice. But potential harm would likely not happen immediately. Any potential benefits from increases in main filter capacitance come with trade-offs as mentioned above. Consider that there is likely some degree of damage to the contacts in your power switch already, and finding a replacement for that switch would be a chore (if doable at all).
There is a triac based modification for power switches that you could look into, as well as other options for increasing lifespan of the switch. I have not done the triac mod myself. And I don't know whether more stout rectifiers are available, or the capabilities of the originals, but you could research. Look at the parameter Ifsm (and others, of course) if you do any research. I have studied lots of rectifier diode spec sheets and have replaced a few sets of those in order to accommodate larger filter caps. But having gone through this process more than once, I am now coming around to an attitude of thinking the benefits are not worth the trouble and long term risks. YMMV
Another thing to consider is that there is a decent chance that any new, large value caps could be at the lower end of their tolerance range for uF. For example, if you find 15k uF caps with +/- 20% tolerance, their actual uF value could be something like 15k -15%, or around 12.75k uF. Can't promise that every case will work out like this, but it does happen enough that it has been discussed by others here (most notably, ConradH), and I have experienced it myself more than once.
Bottom line, if I was doing those amps, I definitely would not install the 22k uF caps. As for the 15k uF caps, I would be researching the rectifiers and also investigating the current health of the power switch (a good thing to do regardless possible filter cap increases). Do you have a means of measuring capacitance values in the range of 15k?