If `io.utf8()` is called while running unit tests, it would be output directly to the console since there is no active file output while testing. Replace call with premake.utf8() as a companion to premake.w() and ignore while capturing to prevent this spurious output.
A number of tests were relying on the fact that x32/x64 were being converted implicitly to x86/x86_64. If/when we retire those symbols, all of those tests would break. Renamed them now to avoid having to sort it out later. Also fixed up comments to keep everything consistent.
