Add updates from Ittiam and Marco.

Bug: aomedia:3281
Change-Id: I5b2ec54744f20a3fc967000466a29498254d9dba

Add updates from Paul and Yunqing.

Bug: aomedia:3281
Change-Id: I7d567d82d86e13170d1240d964bd9c384c95a8bf

Change-Id: Ib0f91cfe2ce757145238a91cd23b754087b49635
(cherry picked from commit 884af216)

Bug: aomedia:3281
Change-Id: If434d027b9476e99b75790e7ac6e8c663a1b00d2

this allows the user to force -DCONFIG_PIC=0 even in the presence of PIE

Change-Id: Id5e3c58542b438cf545f63fa04909c31d42997d5
(cherry picked from commit 1df26009)

this is a follow up to:
905abbb4 av1/arg_defs: use a valid target level index in help msg

the examples are updated to match and the provision to 'pad with leading
zeroes' is removed as it's somewhat misleading and padding the OP index
will make value octal.

Bug: aomedia:3184
Change-Id: I25bf73eda541dde0ca6e98622a9aea9b0a106f26
(cherry picked from commit 6d89b79a)

Fixes the creation of DT_TEXTREL entries in binaries built with PIE
enabled:
  /usr/bin/ld: warning: creating DT_TEXTREL in a PIE

Bug: b/229626362
Change-Id: I37a0eaf6b7471953d2c8b3f7f903c84cb2acbd2d
(cherry picked from commit 7235e657)

Between 3.17.0 and 3.18.2 check_cxx_compiler_flag() sets a normal
variable at parent scope while check_cxx_source_compiles() continues to
set an internal cache variable, so we unset both to avoid the failure /
success state persisting between checks.

Change-Id: I4c9d63d89113d4ce9dc33a937b95df2e9acfce1a
(cherry picked from commit 1c69d568)

previously the index used was 21 = 7.1, which is currently undefined

Bug: aomedia:3184
Change-Id: Idc7fe561ecf25b08c2f88086f7aa2c87ff8c0b58
(cherry picked from commit 905abbb4)

Make the encoder ABI check allow binaries compiled against libaom v3.0.0
or v3.1.x. In those libaom releases, the value of
AOM_ENCODER_ABI_VERSION is 25. Change the encoder ABI check to also
allow the 'ver' parameter to be equal to 25.

Bug: aomedia:3288
Change-Id: I81804be92cbdc77163b415a2156679af64b59520
(cherry picked from commit 45e8b7c56953a5961742d6b7e5f7532333c49d22)

Bug: aomedia:3281
Change-Id: I18a82fe0047b6379d54a8f64674a2af7962b079a

The --auto-intra-tools-off option was added in
https://aomedia-review.googlesource.com/c/aom/+/149021. However, it was
only added as an aomenc option and was not recognized by the key-value
API (the aom_codec_set_option() function). This CL corrects that.

Bug: aomedia:3281
Change-Id: I22933b0388908f31629813d18ee560966c2356f8
(cherry picked from commit 4ab159e7)

Bug: aomedia:3281
Change-Id: I1bd044707704a5ced2a3aa5fedd409ad2b5f8dd8

this avoids a duplicate basename shared with
aom_dsp/arm/highbd_quantize_neon.c

quiets warning in Xcode builds:

XcodeDefault.xctoolchain/usr/bin/libtool: warning same member name
(highbd_quantize_neon.o) in output file used for input files:
AOM.build/Debug/aom_av1_encoder_neon_intrinsics.build/Objects-normal/arm64/highbd_quantize_neon.o
and:
AOM.build/Debug/aom_dsp_encoder_neon_intrinsics.build/Objects-normal/arm64/highbd_quantize_neon.o
(due to use of basename, truncation, blank padding or duplicate input
files)

Change-Id: Id1c2d87c2a1dffca86eaf7d23bb1dcc516082198

Change-Id: Ie1e668f571b5e2bd65ef6c89ca0c0fd03f12efc8

after:
77819417 tune_vmaf: add missing alloc checks

Bug: aomedia:3276
Change-Id: I6c866c0a1d843537fd4af77a1dd344baf09c4958

FPMT related update of framerate and associated timestamp variables
are merged to the default path.

Change-Id: If62901a29c3cc3b138033ee4b72189f9b87f2ca0

Add note about undefined behavior in aom_codec_encode() description.

Bug: b/229104835
Change-Id: I63c5a6f263aa0c6f5c23ebf55e40aacf199d2afc

Bug: aomedia:3276
Change-Id: Icb08c7b9519757e86eca0632f8c7176e37c4c1db

Disabled for now.

Bug: b/221916304

Change-Id: Ibcf606f485d65f0a70cf432b31586700d4bca4ab

Bug: aomedia:3276
Change-Id: I02f7327923a57c89d31c2160068454f696a7bc2a

in tf_alloc_and_reset_data() and av1_apply_temporal_filter_c();
tf_alloc_and_reset_data() now returns a bool

Bug: aomedia:3276
Change-Id: I290416607a21bc3decddbdf14b6dcc0eef337776

in av1_add_to_hash_map_by_row_with_precal_data() and
av1_hash_table_create(); these functions now return bool

Bug: aomedia:3276
Change-Id: Ie7711f3585de9a58567bc1efdcc727007a8cad36

Bug: aomedia:3276
Change-Id: Icc2ac42fa5a35450db22cf038e147e69e694afeb

Bug: aomedia:3276
Change-Id: If39cb118d54576f5ef078df96bf2d6e49c0d49bd

Tried the TODO, but didn't see gains.

Change-Id: I2e9a8b5584d5258b8f97eb6b6992f8b5f069137e

Bug: aomedia:3276
Change-Id: Ie6c0f51d54ba9a0007aa8aca022b462e2cbc54c5

Bug: aomedia:3276
Change-Id: I7d5de84b14451bb7ffbe0a9824ddbb8e38e27c88

Bug: aomedia:3276
Change-Id: I54b7dedead15a2e9bab2532e0733c27635e042e0

Bug: aomedia:3276
Change-Id: I7d555b0eb21a1a99b3ab1132f71e7490e19db0d6

no change in assembly in optimized builds where the functions are
inlined (gcc & clang)

Bug: b/231616924
Change-Id: I5442a7a689d32124d18b1eb7b984806c4628f416

ported from libgav1 @ v0.17.0-83-ge54abf5c

Bug: b/231616924
Change-Id: I07ad128eae7c2a31637aba5da3be17224a8f8abb

ported from libgav1 @ v0.17.0-83-ge54abf5c

Bug: b/231616924
Change-Id: I813d48e386823b457085370cf71776609b405243

to the version based on libgav1 @ v0.17.0-83-ge54abf5c

~2.05-2.34x faster depending on the size

Bug: b/231616924
Change-Id: I9527244a9f6d6a2ecd678fd425f459512cbdb9a3

to the version based on libgav1 @ v0.17.0-83-ge54abf5c

~1.44-1.72x faster depending on the size

Bug: b/231616924
Change-Id: I0ec9bfbdb41d8984594edcd5427542df080c23a9

to the version based on libgav1 @ v0.17.0-83-ge54abf5c

~1.08-1.3x faster depending on the size

Bug: b/231616924
Change-Id: I9c750376ecd217e9252e7450b731a73037bf05dc

paeth/smooth in the 8-bit case

Bug: b/231616924
Change-Id: Id91b8fc380ff1fcc39b403e5329b12d4dea44ed2

In multi-threaded scenario, corrected the
de-allocation of CDEF buffers to be based on
number of allocated workers.

BUG=chromium:1321388

Change-Id: I07579ea4d43875a470beb80316fbdf83a9df47ac

~1.62x faster than previous version.

Bug: b/231719821
Change-Id: Ibb26bee5b3305ea31e7892faf7892b45700b29b6

Change-Id: I449e1e2eed2dd27d25ae20ebbc099fa4275920f1