Took out some ugly case statements and stuck those functions into the asset dictionary instead.  This makes the code a lot more centralized/extensible since if we ever add assets, it'll be clear that their fetch/knowable fields need to be added (versus that logic being in some obscure case statement).

Superficial cleanup to change all typedef std::vector<LLUUID> to use a common typedef uuid_vec_t instead.

This bugs me in a minor way.  Did a search-and-cleanup on various swear words.

and possible fix for EXT-5704: FPS drop, and this could be cause freez and crash.

Checker: FORWARD_NULL
Function: LLWorkerClass::checkWork(bool)
File: /indra/llcommon/llworkerthread.cpp

--HG--
branch : product-engine

by which "moved LLDisplayInfo to llwindow, implemented getting the width/height of screen for mac os and linux."

--HG--
branch : product-engine

Working on major bug EXT-4820([NUX] Viewer dimensions on first-run) - moved LLDisplayInfo to llwindow, implemented getting the width/height of screen for mac os and linux.

--HG--
branch : product-engine

fix for EXT-5683: viewer crashes at llcommon/llworkerthread.cpp(323): ERROR: LLWorkerClass::checkWork: ASSERT(workreq).

--HG--
branch : product-engine

need to specify desctop width and height for macos and linux (see LLDesctopInfo in llsys.cpp for details)

--HG--
branch : product-engine

- new chat source type CHAT_SOURCE_UNKNOWN for avatar names not in cache and object names consisting of two words (avatar names like)
- new icon (unknown_icon.png) for message headers () of nearby chat messagses coming from CHAT_SOURCE_UNKNOWN
- context menu no longer displayed for object names from chat history (no UUIDs)
- double click on message headers in nearby chat no longer opens Inspector (no UUIDs)
- object names (loaded from chat history) in nearby chat are no longer SLURLed (no UUIDs for them)

--HG--
branch : product-engine

Implemented low task EXT-5215 (Disable showing full yyyy/mm/dd timestamp for messages (nearby chat/im) received today)

--HG--
branch : product-engine

Checker: CHECKED_RETURN
Function: LLImageDecodeThread::update(unsigned int)
File: /indra/llimage/llimageworker.cpp

move LLEventTimer into its own source module.  everyone includes it, almost no-one wants it.  now I can dick with it a bit without rebuilding the world, at least. :)

Ugh. Update on Tofu's patch for this bug.
Reviewed by Richard

Checker: UNINIT_CTOR
Function: LLWorkerClass::LLWorkerClass(LLWorkerThread *, const std::basic_string<char, std::char_traits<char>, std::allocator<char>>&)
File: /indra/llcommon/llworkerthread.cpp

Checker: UNINIT_CTOR
Function: LLTreeDFSIter<LLView, std::list<LLView *, std::allocator<LLView *>>::_Const_iterator<(bool)0>>::LLTreeDFSIter()
File: /indra/llcommon/lltreeiterators.h

Checker: UNINIT_CTOR
Function: LLTreeDFSPostIter<LLView, std::list<LLView *, std::allocator<LLView *>>::_Const_iterator<(bool)0>>::LLTreeDFSPostIter()
File: /indra/llcommon/lltreeiterators.h

Checker: UNINIT_CTOR
Function: NamedTimerFactory::NamedTimerFactory()
File: /indra/llcommon/llfasttimer_class.cpp

--HG--
branch : product-engine

--HG--
branch : product-engine

Checker: STRING_OVERFLOW
Function: CProcessor::AnalyzeIntelProcessor()
File: /indra/llcommon/llprocessor.cpp

Checker: STREAM_FORMAT_STATE
Function: LLDate::toStream(std::basic_ostream<char, std::char_traits<char>>&) const
File: /indra/llcommon/lldate.cpp

Checker: STREAM_FORMAT_STATE
Function: encode_character(std::basic_ostream<char, std::char_traits<char>>&, char)
File: /indra/llcommon/lluri.cpp

Checker: RESOURCE_LEAK
Function: LLStringUtilBase<unsigned int>::addCRLF(std::basic_string<unsigned int, std::char_traits<unsigned int>, std::allocator<unsigned int>>&)
File: /indra/llcommon/llstring.h

Checker: INVALIDATE_ITERATOR
Function: LLAllocatorHeapProfile::parse(const std::basic_string<char, std::char_traits<char>, std::allocator<char>>&)
File: /indra/llcommon/llallocator_heap_profile.cpp

not a bug, but make it clearer to coverity.

more reliable fix based on feedback from Richard.  dicked with the Darwin results too since those seemed wrong based on the same feedback (also covered in test plan).

Also define LL_LIKELY/LL_UNLIKELY with a warning about its micro-optimizey nature, and use it to annotate llasserts (i.e. make llassert() lower-overhead and thus more attractive, even in inner-ish loops.)

Gosh, the mac prototypes for get_cpu_clock_count_64 have always been wrong, but the compiler didn't start caring until I made these proper member functions.

fixed.

legacy coding policy fix-up.