Do the content-light statistics x265 computes match the H.265 definition of MaxCLL and MaxFALL?

By reading code against D.3.35, the x265 statistics differ from the spec rule in three ways: they use the Y (luma) plane code values, not per-pixel max(R,G,B) in linear light scaled to cd/m2; the frame average is over the whole stored frame; and the reported MaxFALL is the mean of frame averages, not the maximum. This is a code-derived inference about the statistics, not a statement about what x265 writes in the SEI, which comes from the --max-cll option. No execution was done. Curator qualifications: picyuv.cpp: loop over yPic = m_picOrg[0] (luma plane) computing m_maxLumaLevel = max(yPic[c]) and lumaSum, then m_avgLumaLevel = lumaSum / (m_picHeight * m_picWidth). encoder.cpp 3065-3066: 'm_analyzeAll.m_maxFALL += curFrame->m_fencPic->m_avgLumaLevel; m_analyzeAll.m_maxCLL = X265_MAX(...m_maxLumaLevel)'; 2988-2989: 'stats->maxCLL = m_analyzeAll.m_maxCLL; stats->maxFALL = (uint16_t)(m_analyzeAll.m_maxFALL / m_analyzeAll.m_numPics)'. Curator readback of picyuv.cpp equals producer pin 273fb9c8. Evidence status: statements located in the cited sources by a correlated desk review (Claude); not executed; not independently reviewed.

{
  "citations": [
    {
      "url": "https://www.itu.int/rec/dologin_pub.asp?lang=e&id=T-REC-H.265-202601-I!!PDF-E&type=items",
      "note": "see curator disposition"
    },
    {
      "url": "https://github.com/videolan/x265/tree/419182243fb2e2dfbe91dfc45a51778cf704f849",
      "note": "see curator disposition"
    }
  ]
}