On a new server this function was incorrectly returning "utf8mb4_general_ci".
I started a bug report and it turns out that:
1. The PHP team maintained an internal static list.
2. They ultimately decided to not be bothered to update that internal static list.
So as someone with the intentions on knowing exactly what is going on (e.g. Paris Michigan !== Paris France) I now reject the use of this function.
I'm not sure about other databases though for MariaDB you can determine the many character sets using the following SQL query:
SELECT VARIABLE_NAME, GLOBAL_VALUE, DEFAULT_VALUE
FROM INFORMATION_SCHEMA.SYSTEM_VARIABLES
WHERE VARIABLE_NAME LIKE '%CHAR%';