Use CLCACHE_SINGLEFILE to display actual function calls in cProfile

[Artalus]

This is one possible way to fix #364 .

If you are certain that your build system won't call clcache.exe with multiple source files to compile (AFAIK it is only used by MSBuild to compile sources in parallel with /MP), you can set CLCACHE_SINGLEFILE environment variable to avoid using concurrent.futures.ThreadPoolExecutor and simply call processSingleSource once, avoiding additional process creation. This might increase performance tiny bit, but mostly this is about seing in cProfile results which function calls actually happened and how did they affect the performance.

Another option might be introducing CLCACHE_SINGLETHREAD to simply iterate over file list in a for loop without concurrency. Or maybe even make this decision based on CLCACHE_PROFILE presence.