Improve murmur2 to behave as JS on special chars

Description

The murmur2 code doesn't fit well on special chars (#521)

   $ ./compare.sh 'content: "é"'
+  Hashes do not match
+  native: corivl
+  JavaScript: nawgoz
+  [1]
   $ ./compare.sh 'content: "®"'
+  Hashes do not match
+  native: 3084ue
+  JavaScript: ufik94
+  [1]
   $ ./compare.sh 'content: "😀"'
+  Hashes do not match
+  native: cqlzcw
+  JavaScript: 1u8rdd
+  [1]

How

There were some issues that without handling special char we didn't see before:

JS strings are UTF-16 encoded, so for chars, the max length is 2, but ocaml strings are UTF-8 encoded, meaning they use a variable length (1–4 bytes);
- Then I create the get_utf16_char_codes, and instead of handling with string on the murmur, we hold with a list of utf16 char codes.
The emotion murmur has trick cases, where it handles Int in some instances and Int32 in others. So, I changed to primary Int and used Int32 for some specific cases.

It probably worth a blog post about Int and int32 on JS and utf-16

davesnx / styled-ppx

Improve murmur2 to behave as JS on special chars #522

Description

How