Edit: I started with Footnote 43. I can only find the first page online: http://onlinelibrary.wiley.com/doi/10.1002/j.1477-8696.1982....
Yes, this footnote does say that all entire block of data from 1683 to 1923 is "rather unreliable", and thus your interpretation is correct. But their reasoning for this (only based on the first page) is that for many of the records "no dates of freeze are given". There is no implication that the dates that are given are unreliable, and hence no good reason I can see (from the first page) to eliminate them.
But reading farther, the "red flag" seems more like a "blazing conflagration". In the abstract, they explain that the reason that 1504 to 1515 is considered unreliable is because the written record says that lake didn't freeze for those that 8 consecutive winters. They argue in the abstract that since this is so statistically unlikely, it must be wrong. So this paper is eliminating data as unreliable because of non-freezes, and then concluding that the current string of non-freezes is unprecedented!
Further on the first page is the explicit statement, in complete contradiction to the current paper, that development has indeed affected the freezing of the lake: "In recent years the conditions for complete freezing of the lake have changed because of the increase in the output of the hot spring in the lake. This increase is the result of recreational developments" and "However, the Omiwatari dates are delayed as a result of the recent human intervention". For the new paper to cite this and say there is no evidence that development has affected the timing of the freezing seems (to say the least) disingenuous.