proper float16 etc. support for vertex buffers

VertexBuffer.Format.Half seems to be somewhat of a second-class citizen. I guess VertexBuffer.Format.Double is similarly affected.

• There's no BufferUtils.createFloat16Buffer and most other util functions like setInBuffer are only for FloatBuffers too, ok:

    private static java.nio.ByteBuffer createFloat16Buffer(Vector3f... data) {
        var buf = BufferUtils.createByteBuffer(3 * 2 * data.length);
        for (Vector3f v : data) {
            buf.putShort(FastMath.convertFloatToHalf(v.x));
            buf.putShort(FastMath.convertFloatToHalf(v.y));
            buf.putShort(FastMath.convertFloatToHalf(v.z));
        }
        buf.flip();
        return buf;
    }

    mesh.setBuffer(Type.Position, 3, VertexBuffer.Format.Half, createFloat16Buffer(vertices));

• In many places like Mesh.updateBound or MikkTSpaceImpl a float buffer is assumed for positions/normals/texcoords and it's blindly cast to FloatBuffer which will fail with any other data type. Interestingly for IndexBuffer there's wrapIndexBuffer which deals with the different base types.
• GLRenderer.updateBufferData and convertFormat format handling (GL_HALF_FLOAT etc.)

Start by tracing the FloatBuffer assumptions in Mesh.updateBound and MikkTSpaceImpl, then inspect BufferUtils, GLRenderer.updateBufferData, and convertFormat for Half and Double handling. Done means vertex buffers using non-float formats are supported across these utilities, mesh processing, and renderer format handling.