CRLF end of lines have an origin that comes long before Microsoft (in MSDOS, later Windows, but as well early proprietary versions of Unix made by Microsoft), which just initially used the convention requested and used by IBM and promoted in old standards which which it remained compatible. It dates even before Linux and many Unixes and ad well before Apple's decision to use CR-only newlines.
Those standards are still active and used today. The LF-only convention was a creation on early Unixes and they were proprietary as well when they were introduced to create the C language and *its* initial 'standard' library.
Today, these 3 encodings are endorsed in the standard C library, for which a "text mode" was defined and standardized, and it is formalized in Unicode as well (with a few other encodings whose origin comes also from older IBM standards and EBCDIC, with its use of "C1" controls "LS" and "PS"), and in W3C document formats.
ITU and IETF's BCP's (like MIME) have alway used CRLF as a line terminator as the only conforming convention (which was tentatively extended to allow some freedom, but this never reached the level of becoming used everywhere).
We all have to live with legacy older standards with which we have to remain compatible.
There's been a long life before C, Unix, US-ASCII and the UCS for ISO 10646 and Unicode. Let's not forget it. and note that even Windows no longer requires CRLF newline encodings: the Unicode definition has clarified all this, in full agreement with today's C standard which is independant of Unix/Linux.
