Changes:
Update copyright year to 2010
Updated operator- for Vector3 (use NEG_MASK)
Use newtonrapson_rsqrt4 for Quat::normalize (better precision)
Add Vector::loadXYZW and Quat::loadXYZW, similar to scalar version, using _mm_lddqu_si128
Replace inline by __forceinline
Add get128Ref and set128() methods