parseArgsIO is lazy

Even though it is in the IO monad, `parseArgsIO` does not finish until its result is needed. Unfortunately, this means that argument parsing errors may not happen in time. An obvious workaround is to evaluate some argument test early. Perhaps bang patterns should be added to `parseargs` for this: it is not clear. More consideration is needed.

Start by reading the parseArgsIO entry point and the parseargs implementation named in the issue, focusing on when argument parsing errors are evaluated. Determine an agreed evaluation point and verify that parsing errors occur before the result is needed; the issue does not name a test or file beyond these entry points.

Written by the indexing model from the issue text.