How many tourneys are a good gauge to a player's ability?
Up to how many tourneys can we accept variance for someone's success or losses?

Play around with the variance calculator
https://www.primedope.com/poker-variance-calculator/

evBB/100 with different stack sizes is very good metric and needs a lot less data assuming you don't significantly underperform in ICM and shorthanded game (also if you are very good at winning chips throughout the tournament, but your results are not in expected range according to variance calculator it is likely that leaks are in ICM and shorthanded game).

Some of my ROI stats with cEV info:

I used 0-10bb depth, 10-20, 20-40 and 40+

Site 1: 5bb/100, 4.2bb/100 40bb and under, pretty even numbers with all stacksizes. Result: ~12% ROI without overlays (overlays made it higher)
Site 2: 9.8bb/100, 8.5bb under 40bb stacks, ~30% ROI (highest sample)
Site 3: 17bb/100, 14bb under 40bb stacks ~60% ROI, higher rake no overlays
Site 4: 18bb/100, ~14bb under 40bb stacks ~85% ROI including some overlays

All samples were more than 100k hands and with number of tournaments at least 2* average field size.

E: I have not yet seen professional results without decent chip winrate. There may be some odd player who wins a lot of chips at all stack sizes, but plays poor in final tables to not make money, but I haven't seen such stats with good sample either.

I've no idea what all the parameters mean.