Is simplified writing/accessible tone frowned upon in publications?
It depends on the intended audience.
There is no one rule for writing, academic paper or otherwise. You should always "mold" your content to the level of the average reader of the venue.
Think Nature versus Scientific American, for instance. While I often read SciAm as an undergrad, I didn't really understand most of the Nature articles I had to read recently, because they are dense and full of acronyms that don't mean anything for me. Am I stupid? Maybe, but surely I'm not part of the intended audience of these specific papers. Clearly, both venues are different, one is more technical, the other is focused on scientific dissemination in a broader level. Submitting a very technical paper to a dissemination venue would be a bad idea, and vice versa.
From personal experience, even between technical venues, there is a significant difference. For my research, I alternate between heavily mathematical theory and applications. When I send articles to the IEEE VIS (or TVCG), the reviewers mostly skip the mathematical parts and expect a very "digested", non-mathematical analysis of the results.
I recently sent an article to a more mathematical conference (ISMM), following somewhat the same format I usually use for VIS, with a more detailed mathematical section, but with the same "digested" analysis afterwards. While the mathematical part was accepted mostly as is, the analysis part was heavily criticized, for being "vague" and mathematically inaccurate. It was my fault, I didn't fully mold the content to the venue.
EDIT: Let me more clearly phrase my point: Sometimes reviewers will complain that the paper was too accessible. My example was exactly that, by aiming at a larger audience, writing in "digested" English, instead of harder to understand mathematical definitions, my text was "too vague"/"cryptic".
Something specific to you points:
Simple language: I haven't seen this in any scientific work. I guess it would sound odd but you may have heard of Randall Monroe's book "Thing explainer". I also recall that some time ago there was this short trend where people tried to describe their recent work only using the 1000 most used English words. So, you can do it, but for a journal or conference publication it would be very strange.
Metaphors/Allegories/Similes: Metaphors are in use. I think mainly native speakers and experienced writers use them. But note that metaphors may make it harder for non native speakers to understand the text and also note that some metaphors carry some connotation that not everybody will understand.
Casual language (speaking to the reader, using contractions, etc): I've seen "speaking to the reader" in some textbooks or monographs, but not in papers. For textbooks it makes more sense to me, because you can assume that the reader is indeed a student. For a paper, you generally have no idea if the reader is a grad student, a prof, or a senior scientists at a national lab, so it seems hard to get the tone right. As for contractions: I would avoid them as they don't help and they probably get edited out anyways.
Complex language has a purpose. Journal articles often have length limitations and heavy use of acronyms and complex phrasing can be a very powerful tool for condensing similarly complex ideas into a short article.
Complex phrasing can also have utility in conveying highly precise ideas. Simple language and more accessible writing styles often have to leverage the reader's ability to interpret meaning and context. For casual reading this is fine, but for scientific publication it is typically most important to be extremely precise. Often this requires more dense and complex writing to achieve. This is similar to law, where legal language is equally dense and complex, but to the trained reader leaves little room for interpretation. The meaning is most explicit and unambiguous.
This, similarly, is the primary goal of scientific writing - to leave the audience without need to guess about what has been said. It may not be understood by an inexperienced reader, but when it is understood there is as little room for misinterpretation as possible.