1. JVET-L0271:              Simplification of affine AMVP list construction
2. JVET_L0694/L0045/L0047:  Combination of affine mode clean up and line buffer reduction
3. JVET_L0632/L0142:        Affine merge list construction
4. JVET_L0369:              Moving ATMVP into the affine merge list

Appears to be the source of encoder/decoder mismatches

This is JVET L0054 MMVD code.
MACRO is JVET_L0054_MMVD.

See merge request jvet/VVCSoftware_VTM!26

Squashed commit of the following:

commit d9b44c2820495aee7c77604f511116711a18b880
Author: seungsoo <ss00.jeong@samsung.com>
Date:   Fri Oct 26 08:28:35 2018 +0900

    code update ( reflection )

commit 1fd1b4341134f355651d950ba0024c2b370cdfb5
Author: seungsoo <ss00.jeong@samsung.com>
Date:   Thu Oct 25 10:59:22 2018 +0900

    code update ( reflection of commented things )

commit 3befe26e3e9c23c51bad6e874d868403f665f452
Author: seungsoo <ss00.jeong@samsung.com>
Date:   Wed Oct 24 13:18:27 2018 +0900

    JVET_L0054_MMVD is uploaded.

clean up useless array and its related code; resolve the merge conflict; include MDLM modes in SATD checking.

L0414 (Luma adaptive deblocking filter QP offset) with explit signalling of number of intervals, thresholds, and QPoffsets

integration of JVET-L0338 test5.4.1 and JVET-L0340 test5.6.1

check in the changes of JVET-L0257 on fixing the mismatch between spec and software on the collocated block clipping for the ATMVP

pre-calculating size and position dependent parameters; this change
yields exacatly the same bitstreams

number of regular coded bins", version CE-7.1.3b (with modified
binarization), including restrictions for chroma 2x2 subblocks.