[BUG] `caml_utf8_of_utf16` has a bug in high surrogate case

[bzy-debug]

Describe the bug

caml_utf8_of_utf16(String.fromCharCode(0xdbff, 0xdfff)) generate wrong utf8 encoding string

js_of_ocaml/runtime/js/mlBytes.js

Lines 106 to 113 in 246df64

	} else if (
	c >= 0xdbff ||
	i + 1 === l ||
	(d = s.charCodeAt(i + 1)) < 0xdc00 ||
	d > 0xdfff
	) {
	// Unmatched surrogate pair, replaced by \ufffd (replacement character)
	t += "\xef\xbf\xbd";

line 107 should be c > 0xdbff since 0xdbff is a valid surrogate

Expected behavior

expected result: b'\xf4\x8f\xbf\xbf' (get from python bytes.fromhex('dbffdfff').decode('utf-16be').encode('utf-8'))

actual result: ef bf bd ed bf bf, get from

Array.from(caml_utf8_of_utf16(String.fromCharCode(0xdbff, 0xdfff)))
    .map((c) => c.charCodeAt(0).toString(16).padStart(2, "0"))
    .join(" ")

Versions

latest version of jsoo contains this bug.

[hhugo]

Thanks for the report and proposed fix. See #2008