Ok, so let's go back to your 1997 paper by Yoneda and Watanabe:
Comparisons of altitude tolerance and hypoxia symptoms between nonsmokers and habitual smokers. - PubMed - NCBI
I don't have access to the full paper, but their summary is:
"RESULTS: Smokers revealed significantly fewer subjective symptoms in 5 out of 12 symptoms. There were no significant differences in TUC and the rate of handwriting deterioration between the groups.
Conclusions: Paradoxically, smokers are slightly resistant to hypoxia with respect to emerging subjective symptoms. However, bluntness to hypoxia could postpone the detection of the possible hypoxic occurrence in pilots."
In other words, they didn't find that smokers were any more or less impaired, but they were less skilled at detecting hypoxia i.e. recognising there was a problem. I have to say I don't find handwriting analysis an immediately convincing method of assessing performance.