Complete the change from lldiskcache -> llfilesystem and then addition of new lldiskcache implementation

Rename lldiskcache.* to llfilesystem.* - i think this is the right name since it's responsible for performing file system operations and (will eventually) delegrate to a separate disk cache component to save/load data and keep track of metadata etc.

following promotion of DRTVWR-503

Remove references to VFS in the non-English language XML files (already removed from English (en) versions)

The folder where the disk cache lives was originally renamed from llvfs to llcache but @henri's suggestion that that doesn't reflect the other files in the same place and it should be llfilesystem is a good one so I changed it over

Renamed the references to LLVFile and llvfile.* source code plus cmake scripts to use a different name - lldiskcache - since that more closely resembles what it is (or will be) now that the VFA is no more

First part of change to remove LLVFS from the Viewer. Consists of code changes to remove LLVFS and LLVFSThread classes along with the associated source files. The existing llvfs folder is renamed to llcache. Also includes changes to CMake script in many places to reflect changes. Eventually, llvfile source file and class will be renamed but that is not in this change.

SL-13830 Remade vivox shutdown to terminate when it is told to shutdown instead of when viewer is quiting

Got into D503 by accident

Effective rename of DRTVWR-506-simple since there is no way to rename a branch. Cloned canonical viewer into DRTVWR-519, copied over the files that changed from DRTVWR-506-simple and pushed back.  Once I am satisfied everything is correct, DRTVWR-506-simple will be removed

LN (EKU_TLS) is more likely to be present thus should be checked first

This reverts commit 92499ce1.

# Conflicts:
#	indra/newview/llinventorybridge.cpp
#	indra/newview/llinventorypanel.cpp
#	indra/newview/llinventorypanel.h

This reverts commit 25fd4c2b.

Callstack is clearly broken since it points to LLNotifications::instance().clear(); after 'Goodbye!', my suspicion is that something reinitialized singleton so I fixed cleanup and added some logging to see if there is a dupplicate init

LLWorld is supposed to first remove region from list, then delete the region, but in case of this crash region was returned by handle from llworld yet was mDead and cleaned up, all in main thread.

Coroprosedure should stop on 'stop' exception