Upvote for raising clear points, but at heart I think I disagree with your premises. Your thesis seems to be that by throwing more resources at the problem the vault could be made more secure. Likely true, but at what cost, both in hassle and money?

You seem to be arguing for overengineering without considering the costs. Should the failsafes also have failsafes, should the backups have their own backups? And what if the additional security measures add additional risks? Take #4, "human patrols". Unless implemented very carefully, greater access for security personnel could well add more risk than benefit.

I'm not even sure that one should argue that the vault failed from the point of view of the owners. It was designed well enough for them to do their daily business and such that they were able to obtain insurance on the contents. A vault doesn't have to be impregnable, only visibly difficult to steal from.