Utf8 strings are not null-terminated

[VerySmallRoach]

Tested on 1.91.0.1, but it looks like master have the same issue.

Here's the code:

if (ImGui.Begin("test", ImGuiWindowFlags.AlwaysHorizontalScrollbar)) {
    var label = new string('1', 5000);
    var data = "";
    ImGui.InputText(label, ref data, 10);
}
ImGui.End();

It creates a string with a very long label (to force ImGui.Net code to allocate instead of stackalloc). Sometimes if I launch program and scroll all the way to the right label ends up with random junk changing at runtime:

It does not always happen, sometimes program renders only 1111 as it should, sometimes junk appears for a few seconds and then disappears.

I believe the problem is in these lines (code from decompiled 1.91.0.1):

public unsafe static bool InputText(string label, ref string input, uint maxLength, ImGuiInputTextFlags flags, ImGuiInputTextCallback callback, nint user_data) {
        int byteCount = Encoding.UTF8.GetByteCount(label);
        byte* ptr = ((byteCount <= 2048) ? stackalloc byte[(int)(uint)(byteCount + 1)] : Util.Allocate(byteCount + 1));
        Util.GetUtf8(label, ptr, byteCount);

For long labels this code allocates memory, but is does not seem that Marshal.AllocHGlobal zeroes memory, and Encoding.UTF8.GetBytes does not write null-terminating zero.

Looking at other places in decompiled ImGui.NET code, there's explicit zero after Util.GetUtf8:

public unsafe static bool InputFloat(string label, ref float v, float step) {
        int num = 0;
        byte* ptr;
        if (label != null) {
            num = Encoding.UTF8.GetByteCount(label);
            ptr = ((num <= 2048) ? stackalloc byte[(int)(uint)(num + 1)] : Util.Allocate(num + 1));
            int utf = Util.GetUtf8(label, ptr, num);
            ptr[utf] = 0;
        } else {
            ptr = null;
        }

But InputText and some other methods do not have it.