Summary

Fix releaseWritingBuf() to correctly handle partial writes that split multi-byte UTF-8 characters.

The previous implementation incorrectly converted byte counts to character counts by using:

n = Buffer.from(writingBuf).subarray(0, n).toString().length;

When n bytes cuts through a multi-byte character, the incomplete UTF-8 sequence becomes U+FFFD (replacement character) via .toString(), which has a different .length than the original character. This caused:

• 3-byte characters (CJK) to be silently dropped
• 4-byte characters (emoji) to leave lone surrogates in the buffer

The fix backs up from the byte position to find a valid UTF-8 character boundary by checking for continuation bytes (pattern 10xxxxxx), then decodes the properly-aligned bytes to get the correct character count.

Also fixes a typo where this._asyncDrainScheduled was used instead of the private field this.#asyncDrainScheduled.

Fixes: #61744