Ignore frame type when calculating generic frame dependencies.

When spatial scalability is used, both vpx and aom set key frame flag
for all spatial layers of the first frame, while rtp code expect it to
be set only on the frame without spatial dependencies.

That creates confusion for the frame dependency calculator.
Simplest solution seems to ignore that confusing signal and instead
rely encoder wrappers update frame buffer usages when key frame is generated.

Bug: webrtc:11999
Change-Id: Ica24f1d8d42d32dd24664beabf32ac24872cd15a
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/194002
Reviewed-by: Philip Eliasson <philipel@webrtc.org>
Reviewed-by: Erik Språng <sprang@webrtc.org>
Commit-Queue: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#32667}

modules/video_coding/svc/scalability_structure_test_helpers.cc

@@ -17,7 +17,6 @@
 #include "api/array_view.h"
 #include "api/transport/rtp/dependency_descriptor.h"
 #include "api/video/video_bitrate_allocation.h"
-#include "api/video/video_frame_type.h"
 #include "modules/video_coding/chain_diff_calculator.h"
 #include "modules/video_coding/frame_dependencies_calculator.h"
 #include "modules/video_coding/svc/scalable_video_controller.h"
@@ -56,8 +55,6 @@ void ScalabilityStructureWrapper::GenerateFrames(
       frame_info.chain_diffs =
           chain_diff_calculator_.From(frame_id, frame_info.part_of_chain);
       for (int64_t base_frame_id : frame_deps_calculator_.FromBuffersUsage(
-               is_keyframe ? VideoFrameType::kVideoFrameKey
-                           : VideoFrameType::kVideoFrameDelta,
                frame_id, frame_info.encoder_buffers)) {
         frame_info.frame_diffs.push_back(frame_id - base_frame_id);
       }