Android module changes, stage 1 [defunct]

[ifarbod]

As discussed in #1251, this is a PR that contains merely fixes and minor changes, I'll do the second PR with the new functions and features once this gets merged.

Here's a quick brief of what I did:

• Reworked the stl API, using staticruntime to decide on shared/static.

stl "libc++"
staticruntime "On"
-- Would produce: "c++_static"
-- Or:
stl "gnustl"
staticruntime "Off"
-- Would produce: "gnustl_shared"

• Fixed the MultiProcessorCompilation flag.
• Renamed androidproj to Packaging.
• Prevented some Windows-exclusive attributes from ending up in Android projects, such as CharacterSet.
• Added NEON support for vectorextensions.
• Added UnwindTables support for exceptionhandling.
• Fixed targetName not working in Packaging projects.
• Included Android fixes from bliz #1112 (minus the unit tests) which fixes disablewarnings, exceptionhandling and rtti.
• Unify cppdialect and cdialect, added support for C++latest which maps to C++17 at the moment.
• Fixed pic.
• Fixed thumbmode.

[ifarbod]

Looks like some of the tests are failing, is the order of these really important?

stack traceback:
	...make-core/modules/android/tests/test_android_project.lua:35: in function 'testFunction'
[  FAILED  ] test_android_project.rttiOn
...make-core/modules/android/tests/test_android_project.lua:46: (4) expected:
	<ExceptionHandling>Enabled</ExceptionHandling>
...but was:
	<RuntimeTypeInfo>true</RuntimeTypeInfo>
fulltext:
<ClCompile>
	<PrecompiledHeader>NotUsing</PrecompiledHeader>
	<Optimization>Disabled</Optimization>
	<RuntimeTypeInfo>true</RuntimeTypeInfo>
</ClCompile>

[WorldofBay]

the other wrong test aren't because of order but because of default values in _premake-init.lua

both rtti and exceptionhandling are set to "Default" if not specified.

if the new behaviour is better you need to edit the tests.
i have no idea why the current test suite.exceptionHandlingOff has no rtti set as it should get from the defaults but for the rest the defaults seem to trigger if any configuration is done.

[WorldofBay]

rtti, if not set, defaults to "Default" which will be a true here, while the comment says the default is off, so false.

i don't know which one is correct but either the comment is wrong or you need to add a check for "Default".

also you'll need to change almost every test in the test_android_project suite to look for the correct behaviour for both rtti and exceptionhandling.

[ifarbod]

I believe this bit is from @TurkeyMan's PR.

[WorldofBay]

the rest looks clean but since i know what sam will say i can say it now before he reads it, so you're already done by then:
once you're finished, squash those commits into a single one

[ifarbod]

Thanks for the quick reviews, will do once I figure out the rest of the unit test issues

[ifarbod]

@WorldofBay All of the tests pass now. (the GCC builds on Travis didn't start for some reason)
I'll squash my commits into one if everything else is okay.

[ifarbod]

I messed up my local checkout beyond repair after squashing with premake-core's changes from master, I don't like having a lot of merge commits so here's another PR: #1264