This is going to be an expensive proposition for Tesla. Assuming the glass cracks on 25-30% of cars in the United States where is regular snowfall. In 2018 alone, Tesla sold 138,000 Model 3 vehicles in the US alone, of which ~40,000 may have glass problems (according to my highly scientific aerial extraction.) At the cost of $465 for the glass and $400 in labor costs (another extraction), plus customer transportation costs we're looking at ~$1000 of losses per vehicle for a total of $40,000,000 USD.
This may not seem like a lot in the grand scheme of things, but if you figure into this cost the amount of negative press regarding glass problems (Consumer Reports + Existing Customers) that could have been better directed to positive press and activity on social media (both Tesla and Customer Generated) = additional missed opportunity and sales (with only 1000 lost Model 3 sales) that might have otherwise gone Tesla which easily translates into another $40-60M in losses.
This is a costly mistake that could have cost Tesla a total of 100 million dollars. Seems like doing critical Winter testing would have paid off nicely. Even if Tesla missed this in its analysis, Tesla could have provided (on a temporary basis) ten to twenty cars (free or on a reduced loan) to a handful of testers in the coldest states for feedback.