My question: Is that okay?
Ignoring the given Content-Type and instead guessing what it might be is a bad idea. This can for example result in bypassing external filters (like firewalls or proxies) which rely on the Content-Type for the type of analysis they do. And it would not be the first time if some parts of the browser use the original Content-Type for their decisions while other parts use the guessed Content-Type, which might result in bypassing browser internal restrictions and filters (like XSS protection filter or similar).
Apart from that such unexpected behaver differently from other browsers (and from the standard) is a nightmare for developers and can also cause security problems. Just imagine that the site offers a ways to upload various files which then get offered for download as application/octet-stream, in the hope that this will result in a download. Instead the iOS browser might interpret the file as HTML+JS which essentially results in a stored XSS.
It is recommended to forcible disable such guesses by setting
X-Content-Type-Options=nosniff" in the HTTP header.