Add endianness-aware BinaryReader/BinaryWriter extensions; bump PolyShim to 2.12.1 - #91

PowerKit/Extensions/BinaryReaderExtensions.cs:154

• These float read helpers call BinaryPrimitives.ReadSingle*Endian. This repo targets netstandard2.0/net462 (see PowerKit.csproj), and the PR description notes these float/double BinaryPrimitives overloads aren’t available on all target frameworks (e.g., when using System.Memory 4.6.3). Consider implementing float/double endianness via BitConverter + conditional Array.Reverse so the methods compile consistently across TFMs.

        /// <summary>
        /// Reads a 4-byte floating-point value from the stream using big-endian byte order.
        /// </summary>
        public float ReadSingleBigEndian() =>
            BinaryPrimitives.ReadSingleBigEndian(ReadExactBytes(reader, sizeof(float)));

PowerKit/Extensions/BinaryReaderExtensions.cs:166

• Same concern as ReadSingleEndian: BinaryPrimitives.ReadDoubleEndian isn’t uniformly available across all TFMs this library targets per the PR description. A BitConverter-based implementation avoids conditional compilation and keeps behavior consistent.

        /// <summary>
        /// Reads an 8-byte floating-point value from the stream using big-endian byte order.
        /// </summary>
        public double ReadDoubleBigEndian() =>
            BinaryPrimitives.ReadDoubleBigEndian(ReadExactBytes(reader, sizeof(double)));

PowerKit/Extensions/BinaryWriterExtensions.cs:165

• These float write helpers call BinaryPrimitives.WriteSingle*Endian, but the PR description indicates those overloads aren’t available on all target frameworks (netstandard2.0/net462 with System.Memory 4.6.3). Using BitConverter + conditional Array.Reverse avoids potential multi-target build failures and keeps the API available everywhere.

        /// <summary>
        /// Writes a 4-byte floating-point value to the stream using big-endian byte order.
        /// </summary>
        public void WriteSingleBigEndian(float value)
        {

PowerKit/Extensions/BinaryWriterExtensions.cs:185

• Same concern as WriteSingleEndian: BinaryPrimitives.WriteDoubleEndian may not exist for all targeted TFMs per the PR description. BitConverter-based conversion keeps these methods available across targets.

        /// <summary>
        /// Writes an 8-byte floating-point value to the stream using big-endian byte order.
        /// </summary>
        public void WriteDoubleBigEndian(double value)
        {