ScaledDataConversion : Fix warning 4244 ( possible loss of data in conversion )

- The constructor for `half` only takes a `float` value, causing an implicit cast and warning 4244 when casting a `double` to `half`. Since this function is only being called for `unsigned` types, we can safely replace `std::numeric_limits::min` with 0.

Author: ericmehl

include/IECore/ScaledDataConversion.inl

@@ -163,7 +163,7 @@ struct ScaledDataConversion<
 	T operator()( F f ) const
 	{
 		BOOST_STATIC_ASSERT( boost::is_unsigned< T >::value );
-		f = std::max<F>( f, (F)(std::numeric_limits<T>::min() ) );
+		f = std::max<F>( f, (F)( 0.0 ) );
 		f = std::min<F>( f, (F)( 1.0 ) );
 		float result = static_cast<float>(f) * std::numeric_limits<T>::max();
 		return static_cast<T>( round( result ) );