HLSL: fix dereferencing when copying split structures with arrays

Implement SPV_KHR_subgroup_vote

SPV: Implement new revision of extension GL_AMD_shader_ballot

When copying split types with mixtures of user variables and buitins,
where the builtins are extracted, there is a parallel structures traversal.
The traversal was not obtaining the derefenced types in the array case.

- Add support for invocation functions with "InclusiveScan" and
  "ExclusiveScan" modes.
- Add support for invocation functions taking int64/uint64/doube/float16
  as inout data types.

doc.cpp: Add capabilities, scope to the opcodes. Add opcode and
capability strings.
GLSL.ext.KHR.h: Add extension
string.
GlslangToSpv.cpp: Fix handling of opcodes to generate
appropriate SPIR-V.
spirv.hpp: Add capability and opcode
enums.
spv.shaderGroupVote.comp.out: Update SPIR-V output for test
shader.

This reduces code duplication in a few ways, and better encapsulates
vector swizzle representation.

HLSL: Improved overload handling

Since EOpMatrixSwizzle is a new op, existing back-ends only work when the
front end first decomposes it to other operations. So far, this is only
being done for simple assignment into matrix swizzles.

This partially addressess issue #670, for when the matrix swizzle
degenerates to a component or column: m[c], m[c][r] (where HLSL
swaps rows and columns for user's view).

An error message is given for the arbitrary cases not covered.

These cases will work for arbitrary use of l-values.

Future work will handle more arbitrary swizzles, which might
not work as arbitrary l-values.

This encapsulates where the string could overflow, removing 40 lines
of fragile code.  It also improves handling of numbers that are too long.

There are a couple of open issues that could related to this function
being more rational (locale dependence, 1.#INF).

add other builtins to interstage IO

Includer interface change to support full include semantics (requires downstream changes)

Add a CMake option to disable compilation of HLSL input support.

Matches existing options.

(Still adding tests: do not commit)

This fixes PR #632 so that:

(a) The 4 PerVertex builtins are added to an interface block for all stages except fragment.

(b) Other builtin qualified variables are added as "loose" linkage members.

(c) Arrayness from the PerVertex builtins is moved to the PerVertex block.

(d) Sometimes, two PerVertex blocks are created, one for in, one for out (e.g, for some GS that
    both reads and writes a Position)

Address several compiler warnings

Reorder initializer fields to match variable declaration order (or vice versa) for several class constructors.

Any previous use would only be for "", which would probably mean changing

    include(...)  -> includeLocal(...)

See comments about includeLocal() being an additional search over
includeSystem(), not a superset search.

This also removed ForbidIncluder, as
 - the message in ForbidIncluder was redundant: error results were
   already returned to the caller, which then gives the error it
   wants to
 - there is a trivial default implementation that a subclass can
   override any subset of (I still like abstract base classes though)
 - trying to get less implementation out of the interface file anyway

This is versus a variety of file-related language, designated, requested, etc.
Will simplify diffs for next commit.

- fixed ParseHelper.cpp newlines (crlf -> lf)
- removed trailing white space in most source files
- fix some spelling issues
- extra blank lines
- tabs to spaces
- replace #include comment about no location

PP: #include: simplify the different paths out of #include.

- some paths didn't release 'res'
- token is always '\n' after proper acceptance of the directive itself,
  so no need to test it, change it to '\n', etc.
- assuming setCurrentColumn(0) is not needed unless there are header tokens,
  but not clear why it is ever needed

Note: much of the simplified code read as if the included header tokens had
actually been processed, versus queued up for processing; maybe that explains
some things.

HLSL: allow destination swizzles when writing RWTexture/RWBuffer

This also makes it match how GLSL handles the same thing.

HLSL: allow type keywords as identifiers, and add half type