Fix macro issue to avoid encoder crash when disable IF_12TAP_SIMD,
SIMD_4x4_12 is relying on IF_12TAP_SIMD so that when IF_12TAP_SIMD is turned off, SIMD_4x4_12 should be off automatically.

spanIpmInfoInter() using pu.Y().x + (x << MIN_CU_LOG2) for coordinate would not be correct if called within subpu loop of BDMVR.

- Integration of the adopted config. (Test 4.6)
- (LFNST8, LFNST16) = (32x64, 32x96)

This was raised in https://vcgit.hhi.fraunhofer.de/ecm/VVCSoftware_VTM/-/issues/2
This issue was found to be a problem on Visual Studio 2019 debug and release mode.

when TM_AMVP is turned off by DMVD flag, encoder crash due to the AMVP
list can exceeds size of 2 with the introduction of non-adjacent AMVP
candidates

Renamed source/Lib/CommonLib/x86/sse42/BilateralFilater_sse42.cpp -> source/Lib/CommonLib/x86/sse42/BilateralFilter_sse42.cpp
Changed SIMD-version of the coded so that it also accepts blocks of width 4.

Christopher Hollmann authored 3 years ago and Vadim Seregin committed 3 years ago

# Conflicts:
#	source/Lib/CommonLib/TypeDef.h