track declared builtin type

Marking as WIP since it might deserve discussion or at least explicit consideration.

During type sanitization, the TQualifier's TBuiltInVariable type is lost.  However,
sometimes it's needed downstream.  There were already two methods in use to track
it through to places it was needed: one in the TParameter, and one in a map in the
HlslParseContext used for structured buffers.

The latter was going to be insufficient when SB types with counters are passed to
user functions, and it's proving awkward to track the data to where it's needed.
This PR holds a proposal: track the original declared builtin type in the TType,
so it's trivially available where needed.

This lets the other two mechanisms be removed (and they are in this PR).  There's a
side benefit of not losing certain types of information before the reflection interface.

This PR is only that proposal, so it changes no test results.  If it's acceptable,
I'll use it for the last piece of SB counter functionality.

HLSL: add .mips[][] operator for texture types

This implements mytex.mips[mip][coord] for texture types.  There is
some error testing, but not comprehensive.  The constructs can be
nested, e.g in this case the inner .mips is parsed before the completion
of the outer [][] operator.

   tx.mips[tx.mips[a][b].x][c]

Check Android build on Travis CI

Also stop requiring sudo privilege so that we can use containers
for testing.

HLSL: allow GS-specific methods in non-GS stages

Using GS methods such as Append() in non-GS stages should be ignored, but was
creating errors due to the lack of a stream output symbol for the non-GS stage.

Use unordered_set in ReadableOrderTraverser.

Previously we are using unordered_map to map to bool, which can
be simplified with unordered_set.

Fix for lumped/unity/batch builds

HLSL: Manually configure descriptor set and binding number for resources

The codebase seems to use both “#pragma once“ approach, and “#ifndef / #define” approach,
so I picked pragma once as that one is less typing & less brittle.

When glslang is built with some other build system and lumped/unity builds are used,
without the include guards some headers would get included multiple times, leading to duplicate
declaration errors.

When glslang is built with some other build system and lumped/unity builds are used,
without the checks this would get “macro is being redefined” warnings/errors.

Included changing related parameters to const.
Should fix complaints with previous commit. See issue #879.

Fixes Vulkan Khronos-internal issue 770.

HLSL: implement [unroll] and [loop] attributes

This adds infrastructure suitable for any front end to create SPIR-V loop
control flags.  The only current front end doing so is HLSL.

[unroll] turns into spv::LoopControlUnrollMask
[loop] turns into spv::LoopControlDontUnrollMask
no specification means spv::LoopControlMaskNone

Fix iomapper build warnings

HLSL: Implement missing GatherCmp method

Modernise CMake #2

This reverts commit cfc69d95.
* Change CMAKE_INSTALL_PREFIX default on Windows in order
  to prevent permission denied errors when trying to install
  to "Program Files".

Revert "Modernise CMake"

Modernise CMake

* Use `GNUInstallDirs` in order to respect GNU conventions.
  This is especially important for multi-arch/multi-lib setups.
* Specify position independent mode building properly, without
  using the historic hack of adding `-fPIC` as a definition.
  This makes the build system more portable.
* Only detect C++ (and not C) to slightly speed up configuring.
* Specify C++11 mode using modern CMake idioms.
* Fix some whitespace issues.

Notification phase for io remapper

HLSL: Allow empty struct initializers

Adds a notification phase to the io remapper.
The idea behind this is to give the user a
chance to group uniforms and/or in/out variables
for a better pipeline layout sharing for vulkan.

Change-Id: I7492421085a4156ed3534f01d906ab390d73a623

Review request: does this harm GLSL in any way?

HLSL: add standard sample position return form of GetSamplePosition m…