fix: lxfml normalization (#1835)

dCommon/GeneralUtils.h

@@ -3,6 +3,7 @@
 // C++
 #include <charconv>
 #include <cstdint>
+#include <cmath> 
 #include <ctime>
 #include <functional>
 #include <optional>
@@ -145,7 +146,7 @@ namespace GeneralUtils {
 	template <typename... Bases>
 	struct overload : Bases... {
 		using is_transparent = void;
-		using Bases::operator() ... ;
+		using Bases::operator() ...;
 	};
 
 	struct char_pointer_hash {
@@ -202,7 +203,7 @@ namespace GeneralUtils {
 	}
 
 	template<typename T>
-	requires(!Numeric<T>)
+		requires(!Numeric<T>)
 	[[nodiscard]] std::optional<T> TryParse(std::string_view str);
 
 #if !(__GNUC__ >= 11 || _MSC_VER >= 1924)
@@ -221,7 +222,7 @@ namespace GeneralUtils {
 	*/
 	template <std::floating_point T>
 	[[nodiscard]] std::optional<T> TryParse(std::string_view str) noexcept
-	try {
+		try {
 		while (!str.empty() && std::isspace(str.front())) str.remove_prefix(1);
 
 		size_t parseNum;
@@ -323,4 +324,28 @@ namespace GeneralUtils {
 
 		return GenerateRandomNumber<T>(std::numeric_limits<T>::min(), std::numeric_limits<T>::max());
 	}
+
+	// https://www.quora.com/How-do-you-round-to-specific-increments-like-0-5-in-C
+	// Rounds to the nearest floating point value specified.
+	template <typename T, std::enable_if_t<std::is_floating_point_v<T>, int> = 0>
+	T RountToNearestEven(const T value, const T modulus) {
+		const auto modulo = std::fmod(value, modulus);
+		const auto abs_modulo_2 = std::abs(modulo * 2);
+		const auto abs_modulus = std::abs(modulus);
+
+		bool round_away_from_zero = false;
+		if (abs_modulo_2 > abs_modulus) {
+			round_away_from_zero = true;
+		} else if (abs_modulo_2 == abs_modulus) {
+			const auto trunc_quot = std::floor(std::abs(value / modulus));
+			const auto odd = std::fmod(trunc_quot, T{ 2 }) != 0;
+			round_away_from_zero = odd;
+		}
+
+		if (round_away_from_zero) {
+			return value + (std::copysign(modulus, value) - modulo);
+		} else {
+			return value - modulo;
+		}
+	}
 }