I'll try to explain the problem from both perspectives: author and a journal typesetter.
The typesetting process goes as follows:
We pre-plan the issue contents 2 months in advance, in order to balance the issues in size. This is necessary for small journals with 4 or 6 issues per year, not quite for large journals with a long publishing queue. At this moment, we take articles that are accepted. If there's not enough of them, we go through the queue and try to find articles that can be accepted quickly.
Now the authors provide the final version. This takes some time, so I receive the articles usually 4-6 weeks before the issue date. That's not a lot of time.
Most articles are typeset within 1-2 weeks after I receive them. With these, there's no problem at all. However, then you have articles that take more time, since the quality of the figures is being discussed, as well as semantics (when the formatting from the authors is poor and the semantics are not clear) etc. This takes some time. So it can happen that the article is typeset like 2 weeks before the issue date, or even less.
So now the article is typeset and is with the authors for proofs. Any correction they make has to be incorporated. Sometimes it's not easy (requests for replacing a figure with a better one, for moving figures to other pages etc. are not uncommon). Sometimes I strongly disagree with the authors on these. In such cases, we need to have yet another couple mails exchanged or the chief editor involved, and that takes time again. At this moment you see that 24 or 48 hours can be the maximum we can give.
Once all articles get back, the issue has to be made ready, articles published online, CrossRef+Scopus metadata prepared, DOI registered etc.
That's the perspective of the journal I typeset. I hope that it is clear that the publication comprises a lot of steps. When the authors are cooperative and reasonable, everything goes fluently and the final version is ready 4 weeks before deadline. And then you have cases when things don't go quite well, and you get very close to the deadlines.
Moreover, to make things easier (and reduce the amount of work just before the issue date), you leave authors quite a short time for response. In most cases, there is plenty of time left, but if 80% of people misuse this time, we work 16 hours a day the last 3 days before the issue date to sort everything out, and we simply want to avoid this.
From the perspective of the author, 48 hours is not much for proofreading an article, especially since this has to be done very carefully. However, in most cases, if you ask for extension (a 5-line mail with a very short request is enough), it will be granted without any problem. Just please don't misuse the possibility.