Do I think they should be or should feel obligated to do so? I'm not sure. I think society is having a debate about whether modern communication networks have "public townsquare"-type characteristics and need to be regulated more in general, and this is probably a question on that spectrum to a degree. Let's say I wouldn't be surprised if one day we come to a conclusion that the answer is yes and a larger functionality scope for what is considered basic interoperability is important, but there are also some good arguments against it.
What guides my thinking here is a bit overboard and flowery again, and surprisingly personal. As a software engineer and teenager myself I cut my teeth on making and maintaining one of the better-known IRC clients at the time; this was my first project that really had other users project part of their lives into it. For me it was quite powerful when I realized that I knew people who had met using my software and later gotten married. It solidified my view of the engineer as a toolmaker for others to live their lives and build civilization. The modern equivalent of the people who conceived of and made the flint tools others would use to start the fires to tell stories around. Comm tools being broadly available and interoperable and their social effects are stuff dear to my heart.
Thank you, that's the walled garden perspective I was missing, same as the complaint of those that want to run macOS on non-Mac hw. Usually the walled garden complaints are from the other side of the wall, Apple device users being prevented from installing what they want unless it is already available from AppStore. I didn't consider the other side of that wall.
For example, speaking personally I have no gripe if Apple doesn't want to write an iMessage app for Android, but I think it would be great if the specs and/or APIs were available so that others could write one, even without reaching for one of the regulation-style arguments mentioned above.
Both imply asking Apple to spend effort. With putting macOS on PCs the effort is obvious. But making it possible for others to write an iMessage client also requires effort: For example designing iMessage's technology in such a way that a malicious client can't compromise the system, keeping the protocol relatively stable and sanely versioned, documentation, and so on. With my engineer's hat on, as an opinion, I think this type of effort would contribute to make iMessage better tech as it's in line with the nature of a strong comm protocol, while macOS doesn't benefit as clearly from being on other computers.
The main counter-argument to it is the notion that walled-garden messengers have innovated faster on UX than the standardized ones and then you're back to unresolved quibbles about what's user value and so on. I don't think we've found the one true way to better-tool-to-more-people yet.
For myself and my own grievances, I'm not a fan of comm protocols that don't allow an open market of clients and servers to exist and to interoperate. I much prefer using the ones that do, and so I do a lot of my group chatting on Matrix in the social spaces where I can. The "green bubble effect" and who gets to control it adds another to my list of reasons to want the more open systems to be well-made and prosper, although it's not a primary driver.