fix CHECK on cu mode, considering palette mode

With the decision "Adopt to mandate the presence of AU delimiter
for each AU", AUDs are enabled by default and a warning is printed, if
they are disabled. The config parameter should be removed later and the
decoder needs to check for AUD presence.

SIMD code rewritten to run 2x faster

Clean up virtual boundary related checks:
- Replace modulo operation by bitwise AND
- Define local variable instead of having same computation appear
  dozens of times in code
- Use const type qualifiter where appropriate

following condition is added to fix the bug.
"affMrgType.GBiIdx[affMrgType.numValidMergeCand] = (dir == 3) ? gbiIdx : GBI_DEFAULT;"

The sanitizer reports that a call to getBrickIdxBsMap() tries
to access an out-of-bounds element. The size of the array
brickIdxBsMap[] is thus increased by 1 and its last element is
initialized to some large value

Read 32 bits instead of 64 bits for blocks of width 2 such as to
not read past array boundaries

Addresses a compilation error reported by gcc 9.1

Remy Foray authored 5 years ago and Frank Bossen committed 5 years ago

There is no scaling matrix for palette prediction

  - Add back support for reference wraparound with DMVR.

- refactoring of MatrixIntraPrediction.h/.cpp (merged classes MatrixIntraPrediction and PredictorMIP, removed namespace Mip)
- removed unused variable posBias in function computeReducedPred
- fixed list size in function reduceHadCandList

  - Prevent weighted prediction related processing when IBC or triangle prediction is enabled for the PU.

  - Reduce DMVR SAD cost by 2x to match spec.
  - Adjust BDOF-Disble and Zero-Cost thresholds accordingly.

This part was originally integrated in HEVC SCM8.0 but was later removed as a bugfix in SCM8.3.....

BugFix of encoder deblocking optimization: add initialization of m_ctuXLumaSamples and m_ctuYLumaSamples, align with 4x4 deblocking