BugID: 973

Taking the conjugate of the beamformer weight before sending it to the hardware
was missing. This is needed because:

  //
  // conjugate weights to get the correct matrix

  // weight a_r + i . a_i
  //
  // (a_r + i . a_i) . (x_r + i . x_i) = ( a_r . x_r - a_i . x_i ) + i ( a_i . x_r + a_r . x_i )
  //
  // This is equal to:
  //
  // ( x_out_r )   ( a_r  -a_i )   ( x_in_r )
  // (         ) = (           ) * (        )
  // ( x_out_i )   ( a_i   a_r )   ( x_in_i )
  //
  // and the same for y
  //
  // To send to correct weights to the beamformer we must therefore
  // take the conjugate of the weight (a_r, a_i) to produce (a_r, -a_i)
  // for the x_our_r component and take the conjugate AND multiply by i
  // the weight (a_r, a_i) to produce (a_i, a_r).
  //