Either way, I bet many programmers would run into trouble if asked to drop everything and immediately spin back up a production-ready environment for a project they haven't touched in a year.
I would argue that the whole point of having a publisher is to avoid having to do this kind of thing yourself, but I know many authors forego publishers because publishers don't want to publish their stuff or they don't agree with the publisher's cut. Both reasonable objections.
It's sort of like if you own a restaurant. If someone dies in your family, and at the same time a health inspection gets failed, you have one of two options: Drop everything and coordinate passing the inspection; or deal with your family stuff first and lose income when your product is not selling in the mean time. The more employees and structure you have, the easiest it is to get away for a few days. The more it's a sole proprietorship, the more your fate is in your hands.
(FWIW I am from Newfoundland -- although I'm not an author -- and while the isolation of rural areas is not exaggerated in the comment, it's also not a surprise to people traveling there. When I travel home, which in the last few years has just been for funerals, I have a plan in place)
It’s not as though this is a result of self-publishing.
That is not uncommon, but neither is it the quantitative norm. A lot of authors are particular about the typesetting of their documents, and it does require a rendering environment.
It isn't unusual for a half-LaTeX half-something-else monstrosity to be rendered in a semi-custom environment. Especially if you're dealing with anything academic.
In those kinds of environments you may have a ton of assets, like graphs, that may either be large file-sizes, and/or re-generated during the rendering process from various large datasets. Those assets are not small, and may prevent you holding much more than a couple books on your $4 thumb drive.
Should a professional accountant keep copies of all the tax returns they prepared, on a flash drive that’s with them at all times, in case a client gets audited? Should I keep a copy of every line of code I ever wrote? Where does this professional responsibility end for you?
Accountants are required by law to maintain a minimum of seven years' worth of records for exactly that reason.
Fundamentally, if you want to sell a book you've written, no one is going to care more than you. That means that if you're a small, unsupported author (the vast majority of cases), it's on you to take care of the annoyances that come along for the ride. That's the nature of work. That flagging can be used as a form or harassment is a genuine problem- this is an inevitable side-effect of systems designed for pure automation, without a human in the loop to provide feedback. Over-reliance on automation is an issue that goes far beyond the world of publishing. That is orthogonal to the issue of being responsible for one's own creation, in terms of having needed resources on-hand.
I mean, fundamentally, if you're saying your book needs to be on sale 24/7/365 then you need to be standing behind it 24/7/365 or paying a person or organisation to do that for you. If you want to take a break from the business of your book you should be prepared for Amazon to do the same.
Why should the author, rather than the publisher, be expected to produce a public-upload-ready file on demand? (If the deadlines other authors are reporting in the thread are accurate, I wouldn't even reasonably expect a full-time major publisher to hit them - and during the last week of the year?!)
(Well, technically Amazon is the publisher, but in Amazon's lexicon it's self-published.)
Example: https://www.smashwords.com/list
"Self-publishing" is really orthogonal to both of those, though today most people would probably describe any vanity press as effectively self-published.
In this case, (further down the thread) it was a dialect thing, so the whole problem was a false positive from an over zealous automated system on Christmas Eve, in prime panic ebook gifting time.
The majority of authors today don't know that software named pandoc exists, let alone what it does, let alone what it might do for them.
That the automated system as described can be abused is a real problem, and that needs to be addressed. But that is a separate issue from the generally bad state of what passes for final edits in many books nowadays. That the author in question can't be bothered to keep a current copy of their manuscript leaves me lukewarm on the issue.