add 64F intrinsic in HAL NEON

  * use universal intrinsic for accumulate series using float/double
  * accumulate, accumulateSquare, accumulateProduct and accumulateWeighted
  * add v_cvt_f64_high in both SSE/NEON
  * add test for conversion v_cvt_f64_high in test_intrin.cpp
  * improve some existing universal intrinsic by using new instructions in Aarch64
  * add workaround for Android build in intrin_neon.hpp

modules/core/include/opencv2/core/hal/intrin_sse.hpp

@@ -1587,11 +1587,21 @@ inline v_float64x2 v_cvt_f64(const v_int32x4& a)
     return v_float64x2(_mm_cvtepi32_pd(a.val));
 }
 
+inline v_float64x2 v_cvt_f64_high(const v_int32x4& a)
+{
+    return v_float64x2(_mm_cvtepi32_pd(_mm_srli_si128(a.val,8)));
+}
+
 inline v_float64x2 v_cvt_f64(const v_float32x4& a)
 {
     return v_float64x2(_mm_cvtps_pd(a.val));
 }
 
+inline v_float64x2 v_cvt_f64_high(const v_float32x4& a)
+{
+    return v_float64x2(_mm_cvtps_pd(_mm_castsi128_ps(_mm_srli_si128(_mm_castps_si128(a.val),8))));
+}
+
 //! @endcond
 
 }

modules/core/test/test_intrin.cpp

@@ -626,12 +626,18 @@ template<typename R> struct TheTest
         dataA *= 1.1;
         R a = dataA;
         Rt b = v_cvt_f64(a);
+        Rt c = v_cvt_f64_high(a);
         Data<Rt> resB = b;
+        Data<Rt> resC = c;
         int n = std::min<int>(Rt::nlanes, R::nlanes);
         for (int i = 0; i < n; ++i)
         {
             EXPECT_EQ((typename Rt::lane_type)dataA[i], resB[i]);
         }
+        for (int i = 0; i < n; ++i)
+        {
+            EXPECT_EQ((typename Rt::lane_type)dataA[i+n], resC[i]);
+        }
 #endif
         return *this;
     }