General Wrapper Optimizations

[johnwparent]

Optimizes wrapper performance in a few primary ways:

• removes manual stdout/err capture and forwarding
  rather than redirecting the toolchain child processes stdout/err to pipes we read from and write to the wrapper stdout for capture, we directly attach the child to the wrapper's stdout and avoid having to spin a cpu to pipe io streams.
• Uses waitforsingleobject instead of calling GetExitCodeProcess while waiting for the child object to return.
  GetExitCodeProcess within our while loop introduces a busy wait loop vs allowing the scheduler to pause execution entirely, allowing for more concurrent processing and reducing parallel bottlenecks
• Builds with more compiler and linker optimizations
• uses more optimal data structures and types when appropriate
• When debugging, checks for debug before constructing debug string
• Reduces frivolous operations (like checking env variables we don't use)
• Pulls regex utils into its own header so the heavy <regex> header is only included in TUs that actually need it.

Adds a benchmarking test to be run whenever changes to multiprocessing/threading logic are made to ensure continued performant execution of the wrapper.