Here I was hoping that if you took the UTF-8 representation in bytes and decoded it as ASCII, you would get something interesting. But no, just Unicode characters. Almost interesting is that none of the bytes are valid ASCII characters (< 128), which you might expect for the first byte of every UTF-8 codepoint due to backwards compatibility for ASCII encoding, but perhaps not for the subsequent bytes that comprise the rest of the grapheme.
I’m finally starting to understand the appeal of numerology.
Just to verify all permission-related things in one go, see if you can open the key as your user with an editor like vi or nano. This will let you separate out some behavior specific to OpenSSL vs some behavior purely permissions-based.
I’m not sure what’s happening here, but the above test can at least narrow the focus.