You promise stakeholders and customers that a product will be ready to ship by a certain date. These things are very easy to underestimate. So in the end you have lots of extra bugs to fix, and you have a huge amount of tension between releasing the (buggy) software on time, letting the delivery date slip, and putting in whatever it takes to release something of acceptable quality on time.
Moreover pre-hyping releases is an incredibly easy way to get a lot of publicity for your product, so it is very common.
So the fundamental issue is marketing, and everything cascades from there. But even if it isn't a normal part of development at your firm, if it happened because something major was discovered last minute and required a lot of work, would you have considered postponing your paternity leave?
Compare this to biotech. Here you have elaborate licensing and safety studies (of dubious effectiveness these days) which occur between the initial product development and its release to market. If Novo Nordisk comes up with a new drug tomorrow, they can't even estimate a release date because of all the regulatory stuff that comes with it. So this slows the march to production down and prevents death marches.
So I think the issue with death marches is that they happen, not as a function of or product development, but as a function of PR and marketing, which leads to pre-hyping a specified day of release, and if you let it slip too much, you lose out on all that built-up hype.
So I am not saying they should be normal. I am saying they are systemic, and expected particularly in a start-up phase of a business.