Issue in condition to parse lfnst_idx when treeType = DUAL_TREE_CHROMA

[forayr]

This concern the following condition:

  if( cu.ispMode != NOT_INTRA_SUBPARTITIONS || 
#if JVET_O0925_MIP_SIMPLIFICATIONS
      (cu.cs->sps->getUseLFNST() && CU::isIntra(cu) && cu.mipFlag && !allowLfnstWithMip(cu.firstPU->lumaSize())) ||
#else
      cu.mipFlag == true ||
#endif
    ( CS::isDualITree( *cu.cs ) && cu.chType == CHANNEL_TYPE_CHROMA && std::min( cu.blocks[ 1 ].width, cu.blocks[ 1 ].height ) < 4 )
#if JVET_O0213_RESTRICT_LFNST_TO_MAX_TB_SIZE
#if JVET_O0545_MAX_TB_SIGNALLING
    || ( cu.blocks[ 0 ].width > cu.cs->sps->getMaxTbSize() || cu.blocks[ 0 ].height > cu.cs->sps->getMaxTbSize() )
#else
    || ( cu.blocks[ 0 ].width > MAX_TB_SIZEY || cu.blocks[ 0 ].height > MAX_TB_SIZEY )
#endif
#endif
    )
  {
    return;
  }

When CS::isDualITree( *cu.cs ) && cu.chType == CHANNEL_TYPE_CHROMA, ie treeType = DUAL_TREE_CHROMA, the data in cu.blocks[0] are not valid. For exemple, in this case, cu.blocks[ 0 ].width is equal to 0. Information from cu.blocks[1] should be used.

[forayr]

Proposed fix: ​https://vcgit.hhi.fraunhofer.de/jvet/VVCSoftware_VTM/merge_requests/806

[ksuehring]

The fix was merged.