Add multi-thread safety testing for AST linking and SPIR-V generation

Hello,

I'm looking for documentation/examples regarding multi-threaded shader compilation using the C++ API in in order to compile from GLSL -> SPIRV. In the standalone app there is a use case with the legacy API only.

From early tests I didn't have issues, but I wonder what is the correct/intended way to achieve this. And most importantly, whether it is safe to use the new C++ API in such a multi-threaded way.

e.g., something like
```
parallel_for(0, incomingShaderCode.size(), [](int index)
{
... do setup of the shader (single shader)
TShader* shader = new TShader;
shader->setStringsWithLengthsAndNames(incomingShaderCode[index].code, ...)
shader->parse(...);

... create the program and generate SPIRV
glslang::TProgram& program = *new glslang::TProgram;
program.addShader(shader);
glslang::GlslangToSpv(program.getIntermediate(), ...)

... cleanup
}
````

Thank you in advance.

Begin with the C++ API entry points named in the report—TShader, TProgram, and GlslangToSpv—and compare them with the legacy API path used by the standalone app. Define the intended thread-safety behavior for parallel GLSL-to-SPIR-V compilation and add focused multi-thread coverage for AST linking and SPIR-V generation; the issue does not name files or existing tests.

Written by the indexing model from the issue text.