Polyfill vmlaq_lane_f32 for Webassembly

Question

Webassembly doesn't currently support FMA instructions and vmlaq_lane_f32 fails to compile with Emscripten. So far I came up with the code below but there's no such thing as vaddq_lane_f32 and vaddq_f32 works on all lanes so it's not the best solution.

      float32x4_t vmlaq_lane_f32(float32x4_t a, float32x4_t b, float32x4_t c, lane) {
            return vaddq_f32(a, vmulq_lane_f32(b, vget_low_f32(c), lane));
      }

Second issue is lane must be const and the above doesn't work as a standalone function so I have to replace every instance manually.

One more problem is that multiple files should include this definition and the compiler errors out due to redefinition in o. files.