fix incorrect LS conversion matrices

The matrices are stored in a compressed format, while testing it was discovered that the matrix for downmixing 5_1_4 to 5_1_2 is incorrect:

In [1]: old = np.zeros(10*8)

In [2]: old[[0, 11, 22, 33, 44, 55, 66, 77]] = 1.0

In [3]: old[[48, 59]] = 0.89

In [4]: new = np.zeros(10*8)

In [5]: new[[0, 9, 18, 27, 36, 45, 54, 63]] = 1.0

In [6]: new[[68, 77]] = 0.89

In [7]: old.reshape((10,8))
Out[7]:
array([[1.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ],
       [0.  , 0.  , 0.  , 1.  , 0.  , 0.  , 0.  , 0.  ],
       [0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 1.  , 0.  ],
       [0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ],
       [0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ],
       [0.  , 0.  , 0.  , 0.  , 1.  , 0.  , 0.  , 0.  ],
       [0.89, 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 1.  ],
       [0.  , 0.  , 0.  , 0.89, 0.  , 0.  , 0.  , 0.  ],
       [0.  , 0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  ],
       [0.  , 0.  , 0.  , 0.  , 0.  , 1.  , 0.  , 0.  ]])

In [8]: new.reshape((10,8))
Out[8]:
array([[1.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ],
       [0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ],
       [0.  , 0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  ],
       [0.  , 0.  , 0.  , 1.  , 0.  , 0.  , 0.  , 0.  ],
       [0.  , 0.  , 0.  , 0.  , 1.  , 0.  , 0.  , 0.  ],
       [0.  , 0.  , 0.  , 0.  , 0.  , 1.  , 0.  , 0.  ],
       [0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 1.  , 0.  ],
       [0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 1.  ],
       [0.  , 0.  , 0.  , 0.  , 0.89, 0.  , 0.  , 0.  ],
       [0.  , 0.  , 0.  , 0.  , 0.  , 0.89, 0.  , 0.  ]])

The fix is a simple patch with the updated indices in the table for the decoder + similar patch for pyaudio3dtools (already done).

Update 2022.07.18:

I also found that the 7_1 to 7_1_4 conversion was performing a 1:1 upmix, however due to slightly different speaker ordering this was also incorrect and needed an update.