LLControlGroup::loadFromFile() can of course detect which LLControlGroup
instance it's loading. We only want to log unrecognized settings variables in
LLControlGroup "Global". Settings for "Don't show me this again" notifications
are in group "Warnings".

Change LLControlVariable::mPersist from bool to tri-state enum. Its 'true'
value used to mean "persist only if changed from default;" third state now
means "persist regardless of value." Add forcePersist() method to set that.
Replace isSaveValueDefault() method -- used only by logic to determine whether
to save the variable -- with shouldSave() method to encapsulate ALL that logic
rather than only part of it. shouldSave() recognizes PERSIST_ALWAYS state.
When loading an unrecognized control variable from a user settings file, use
forcePersist() to ensure that we later save that variable again.
Tweak one of the unit tests to adjust for new semantics.

LLControlGroup::declareControl(), declareString() etc. etc. all used to return
BOOL -- which no one ever examines because it unconditionally returned TRUE.
Make it return the (possibly new) LLControlVariable* instead.

A small, fixed set of cmd_line.xml switches can't reasonably be mapped to
settings variables, mostly because they affect the settings machinery itself.
Other than those, every new cmd_line.xml switch should map-to a settings
variable. Validate that only the known set does not have map-to; validate that
map-to variable actually exists.

Route --url and --slurl command-line switches through a common settings
variable. Treat them uniformly now. (Previously, passing --url would notice a
grid-specific SLURL and preselect that grid; --slurl wouldn't. Now both do.)

I've never really understood the usefulness of displaying world-global
coordinates in the Help -> About box. It seems to me far more useful to know
where you are within the current region. If that proves problematic, we can
display both sets of coordinates -- but let's try it this way first.

The existing POSITION variable gives "global" position: that is, your region-
local coordinates plus the (somewhat arbitrary) global coordinates of the
region's corner within the whole world. That may be meaningful to people on
the mainland, hard to say, but it correlates with nothing else available from
the viewer. POSITION_LOCAL gives you region-local coordinates, which could be
used (for instance) to construct a SLURL.

The cmd_line.xml entries:
analyzeperformance
crashonstartup
debugsession
disablecrashlogger
logmetrics
logperformance
noquicktime
replaysession
all specify map-to settings.xml variables -- none of which existed! Introduce
such variables. Instead of detecting the presence of a particular switch in
the command-line parser, use the value of its corresponding settings variable.

Use map-to in cmd_line.xml to inform the command-line processor that the
target variable for --graphicslevel is RenderQualityPerformance.
That lets us eliminate clunky llappviewer.cpp switch from '0' to 0, etc.
Moreover, previous switch statement only accepted 0 - 3, whereas
LLFeatureManager::setGraphicsLevel() actually accepts 0 - 6. Introduce
LLFeatureManager::isValidGraphicsLevel() and use that to validate.
Replace switch statement in setGraphicsLevel() mapping int constants to string
literals with static vector of level names, using same data for mapping as for
validating level numbers.

UserLoginInfo block had <key>Value</key> without the required subsequent
element to provide any actual value. This confused at least the Python
LLSD reader.

It seems that certain build hosts have an (obsolete? broken?) install of
indra.util.llmanifest under the system Python. If we append the local repo
indra/lib/python to sys.path, viewer_manifest.py pulls in the broken
llmanifest. Prepend to sys.path instead to ensure we get the right one.

Also clarify comment for ELLPath in lldir.h: ELLPath int values are read from
settings_files.xml.

However, for backwards compatibility, continue to recognize and discard
--skip-gridargs switch.

Instead of generating viewer command-line switch overrides into
the Windows shortcut,
an arguments.txt file on Mac, or
a gridargs.dat file on Linux,
generate a platform-independent settings_install.xml file containing
'sourceid' key (if 'sourceid' found in TeamCity environment variables).
All command-line override switches previously found in the aforementioned
files have happily become moot.

Disable asserts in CalculateTangentArray (discussed with davep) to avoid debugging interruptions from bad assets

MAINT-2632 speculative guards in display_startup based on prevalence of crashes ending there (no repro, unfortunately)