WriteValuePacked Bug

Description

Hello,I find when use the WriteValuePacked api to pack a float value, it will convert the float to uint.Due to the storage format of the float the convert value will bigger than 1 << 29,which can lead to the result of using 5 bytes to serialize the float,so can you tell me the reason why we convert the float to uint.

Reproduce Steps

Use the test runner to pack the float number.

Actual Outcome

Lead to the result of using 5 bytes to serialize the float.

Expected Outcome

Use less than 4 bytes to serialize the float.

Screenshots

從 WriteValuePacked API 開始，使用 test runner 重現 float 情況。檢查 float-to-uint 轉換和 packed serialization 的行為；當確定目前位元組數的原因，並且有一個可重現的測試驗證預期的序列化大小時，即表示完成。