The political situation on Svalbard (including Spitsbergen) is complicated. Norway technically owns the islands and the Norwegian krone is the technical legal tender there, but the Russians have considerable rights to mineral exploitation, which go far beyond the usual foreign-owned mining interests known in other countries. The coal mine workers were Russian, and the company would have wanted to pay them in a monetary form that they were both familiar with and could easily convert into funds for sending back home to their families. The Soviets would also have been reluctant to pay their workers in hard Norwegian currency, in case any of them got bright ideas about defecting with it. But I suspect there may have been political pressure from the Norwegians not to allow stockpiles of actual Soviet/Russian currency onto the islands. The mining token coinage was presumably a compromise acceptable to both sides.
If this is true, then it's highly likely that the periods when no tokens were issued were periods when Soviet/Russian money was being allowed to be imported onto the islands and used instead.
Then, of course, one would also draw comparisons between the Spitsbergen mining tokens and the various mining token series issued by Denmark for mining colonies on nearby Greenland, where none of the issues mentioned above came into play yet mining tokens were nevertheless deemed necessary.
As for the "legitimacy" of the three series, I'd count 1946 and 1993 as legitimate circulation coins; 1998 are not. This seems to be the conclusion of Krause as well; the 1946 and 1993 tokens are listed in the mainstream Krauses, but the 1998 series are now listed in the "Unusual World Coins" book and given KMX numbers.
Don't say "infinitely" when you mean "very"; otherwise, you'll have no word left when you want to talk about something really infinite. - C. S. Lewis