Add a clang tool to detect std::move(raw ptr)
The CL also adds a sentence to the documentation for clang tools, to supply a
bit which was unclear for me during writing of the added tool.
So far, on Linux, the tool found instances of std::move(raw ptr) in:
../../components/browser_sync/profile_sync_service.cc:10101
../../components/sync_sessions/sessions_sync_manager.cc:5987
../../content/browser/indexed_db/indexed_db_transaction.cc:3865
../../gpu/ipc/service/gpu_channel.cc:26467
../../services/preferences/tracked/pref_hash_store_impl.cc:3329
../../services/preferences/tracked/pref_hash_store_impl.cc:4518
../../services/ui/public/cpp/bitmap/child_shared_bitmap_manager.cc:5009
../../services/video_capture/test/mock_device_factory.cc:2266
It also found it in some 3rd party headers, such as
../../third_party/flatbuffers/src/include/flatbuffers/flatbuffers.h:8260
../../third_party/skia/include/core/SkCanvas.h:14756
../../third_party/webrtc/api/proxy.h:16937
../../third_party/webrtc/api/proxy.h:19681
../../third_party/webrtc/api/proxy.h:20241
and c++ libs.
The Chromium 1st party instances will be addressed in separate CLs. For the headers, I'll consider if it makes sense to address this (e.g., some occurrences seem to be template instantiations, where the template is not a priori aware that the supplied type is a raw pointer; in such cases std::move should stay).
I may also run the tool on more platforms.
Finally, while preventing regressions would be good, it seems impractical to run this tool as a presubmit check. However, I will try to run it after https://crbug.com/646113 is finished, because that base::Value refactoring might introduce some instances (happened to me recently).
BUG=729393,731577
Review-Url: https://codereview.chromium.org/2919243002
Cr-Commit-Position: refs/heads/master@{#478226}
5 files changed
