Yes, you can attain statistical significance with large sample (N > 100,000). No, it's still insignifcant, because effect size is so small. Did you read the article? The effect is less than 1 point of IQ, whereas black-white IQ gap is usually quoted as 15 points.

Hmm? They find that one additional year of schooling increases IQ by 3.7 points. I'd say taking that figure with the difference in schools that blacks and whites in the US usually attend and you've got a plausible answer that the IQ gap is created through environmental differences.

I think the PISA tests disprove your hypothesis.