HLSL: fix several issues in mat construction from scalars

This fixes:

1. A compilation error when assigning scalars to matricies

2. A semantic error in matrix construction from scalars.  This was
initializing the diagonal, where HLSL semantics require the scalar be
replicated to every matrix element.

3. Functions accepting mats can be called with scalars, which will
be shape-converted to the matrix type.  This was previously failing
to match the function signature.

NOTE: this does not yet handle complex scalars (a function call,
say) used to construct matricies.  That'll be added when the
node replicator service is available.  For now, there's an assert.

There's one new test (hlsl.scalar2matrix.frag).  An existing test
lsl.type.half.frag changes, because of (2) above, and a negative
test error message changes due to (3) above.

Fixes #923.

SPV: Fix an typo of SPV_AMD_texture_gather_bias_lod

E.g., in

    float float;
    (float) * float;

The "(float)" is not a type cast, it is an expression.

HLSL: compilation warning fix: no functional change

One liner to eliminate a compile warning.

- Add int16 types (int16_t, uint16_t, i16vec, u16vec).
- Add int16 support to GLSL operators.
- Add int16 type conversions (to int16, from int16).
- Add int16 built-in functions.

This implements Khronos-private specification bug 16130.

This tracks Khronos-private specification bug 15945.

Infrastructure: add include search paths (-I, etc.).
Fixes #921.

This lets all languages share the same definitions.

Build: Fix linux

I believe this is a repeat of 12d69368.

Unlike "if (XXX)" and "while (XXX)", with "for (YYY...", the YYY can be
more kinds of statements than the XXX.

Implement the extension GL_AMD_texture_gather_bias_lod

Setting component=0 is later taken to mean the shader did so, which
is not always legal. It should instead set the component as not set
by the shader.

This implements a recent change to the GLSL specification to enforce
this ill-defined situation.

Remove empty cpp files

HLSL: add test coverage for sub-vec4 texture intrinsics

SPV: When passing structs of opaque types, flatten and pass the membe…

For "s.m = t", a sampler member assigned a sampler, make t an alias
for s.m, and when s.m is flattened, it will flatten to the alias t.
Normally, assignments to samplers are disallowed.

This avoids either A) needing uniformConstant struct, or
B) initializing a struct with opaque members, as writing them is not
allowed.

Because it is valid in HLSL to alias bindings:
A) remove validation that aliasing is not done
B) make the algorithms tolerate aliasing

This changes no functional code.  There was a bit of a testing hole
in that textures templatized on sub-vec4 types were not being exercised
with any intrinsics.  This adds some basic sanity coverage of that case.

HLSL: Add an Includer to handle #include for local HLSL paths.