The N35/N42/N50/etc. refers to the grade of the magnets, which corresponds to the maximum possible magnetisation. Basically: larger number = stronger magnets. "4 × 2" refers to the size in millimetres (first number is the diameter, second number is the thickness).
I don't have any input regarding what magnets to use since I haven't really tried either the Valk Power M or the Aochuang GTSM, but you can check out teh yoshi's spreadsheet for a general guide.
More expensive and harder to mass produce, like others have said, but the key thing is that the magnetic pull when the layers are aligned is not the only thing that affects how magnets feel in a puzzle. Being a bit stronger when aligned wouldn't mean much if the slightest misalignment also causes the force to fall off steeply—this wouldn't be very different from classical friction-based alignment mechanisms, for example. For small magnets, the force between them obey an approximate inverse fourth power law, not inverse square (because the magnets are dipoles, not monopoles). You can't make the magnets too small for that reason, even if they're very strong. (Disclaimer: I haven't done the calculations (which can get very complicated), but "very small" here probably means something like 1 mm diameter.)
That said, a few puzzles feature magnet grooves where the plastic is thinner (e.g. Galaxy v2 and I believe the Aoshi GTS M as well), so it's not like puzzle designers haven't been thinking about reducing magnet distance.
