ICr suggests as many factors as variables, despite a very strong first component

[ffabio-econ]

The issue is explained in details here: https://stats.stackexchange.com/questions/657587/weird-results-from-bai-ng-pcs-selection-criteria-implementation-of-dfms-on-r

As the title suggests, with some temperatures data I get that I should keep as many components as variables, despite a first principal component explaining around 80% of the variance... I don't understand where the problem comes from.