On something so absolute as Ethereum, I would expect them to use a very dumb language with as few complications and complexities as possible, and then to formally verify the crap out of it, and quickly make it the most consistent, predictable, and stable language on the planet (again, possible only with a very barebones language that does not have complex features). The fact that Ethereum has been around for so many years and this has not happened indicates that it's not a serious platform.
What other classes of errors are there that Ethereum is successfully avoiding?
Well, that's reassuring. Should we expect that there will be no more programmer errors in the future on the Ethereum network then? Isn't Ethereum's job to minimize those errors?
Most of the largest headlines involving loss of funds in the Ethereum network have been due to programmer or user error. Keep in mind that Ethereum is still in its "experimental" phase - it isn't "supposed" to have billions of dollars put into it just yet, but the market doesn't care about this as long as the prices are going up. Ethereum could be much worse than it is, but that wouldn't mean its value would go to zero, or even do down. Plenty of top coins by market capitalization have been orders of magnitude worse, or just outright scams. Remember "The market can stay irrational for longer than you can remain solvent".
It's still a very new area of technology despite the amount of money that pours into it. There's innovation, but it can be difficult to see it in the midst of the hype.