Fix buffer overflow in 64x16 ssse3 idct

With frame threading enabled the code could previously clobber the
coefficients of the next block.

Update the checkasm test to check for this.
22 jobs for master in 6 minutes and 41 seconds (queued for 3 seconds)
Status Job ID Name Coverage
  Style
passed #276752
amd64 debian
style-check

00:00:22

 
  Build
passed #276753
amd64 debian
build-debian

00:00:41

passed #276761
debian aarch64
build-debian-aarch64

00:01:32

passed #276762
debian aarch64
build-debian-aarch64-clang-5

00:01:02

passed #276765
debian armv7
build-debian-armv7

00:02:43

passed #276766
debian armv7
build-debian-armv7-clang-5

00:01:10

passed #276754
amd64 debian
build-debian-static

00:00:39

passed #276764
debian aarch64
build-debian-werror

00:00:34

passed #276755
amd64 debian
build-debian32

00:00:39

passed #276763
macos
build-macos

00:00:28

passed #276767
amd64 debian allowed to fail
build-ubuntu-snap

00:01:04

passed #276759
amd64 debian
build-win-arm32

00:00:26

passed #276760
amd64 debian
build-win-arm64

00:00:31

passed #276756
amd64 debian
build-win32

00:00:43

passed #276757
amd64 debian
build-win32-unaligned-stack

00:00:40

passed #276758
amd64 debian
build-win64

00:00:48

 
  Test
passed #276768
amd64 debian
test-debian

00:00:45

passed #276770
amd64 debian
test-debian-asan

00:02:15

passed #276771
amd64 debian
test-debian-msan

00:01:08

passed #276772
amd64 debian
test-debian-ubsan

00:01:33

passed #276769
amd64 debian
test-debian-unaligned-stack

00:00:43

passed #276773
amd64 debian
test-win64

00:00:59