comments for non-ascii character is inaccurate

[bobzhang]

'文'-> Js.log

Generated JS output is

console.log(/* '\391' */25991);

[butterunderflow]

This issue seems raised from here:
https://github.com/rescript-lang/rescript-compiler/blob/54e6a2bb061a61b34d8a9a2b1a37f68d1942f26d/jscomp/core/js_dump.ml#L633

When a char with a big encode value is passed to here(like '文', with encode 25991), format by %C will not get the encoded value back(given '\391' here).
Changing the payload char to int all the time, and providing a safer string_of_int_as_char-like function seems to solve this.
I'm working on this.