We've seen this scenario before: performing Kolmogorov-Smirnov tests, if it seems to materialize, is actually just a conditional step and not a decisive proof of the validity of a methodology.
Historically, data quality and contextual relevance have always been more important than mechanical application of a test.
If the data used reflect abnormal market periods, like Japan's zero-yield policy, the test result could be misleading even if it indicates similarity.
For example, a KS test on Japanese asset performance data might seem valid but would ignore the distorting impact of the massive quantitative easing by the Bank of Japan on return distributions.