Extend build.sh to normalize test output for test comparison, regression identification

[IgorTodorovskiIBM]

Idea:

• Normalize the test output of various tools so that we can do actual/expected comparisons of each tests
• Jenkins can use a generic logic to identify regressions, report test failures

Implementation idea:

• build.sh will offload the parsing to a tool specific script like portparsetests.sh which will parse it into 3 fields (testcase, pass/fail, fail message) and store it into a csv or other standard file format (junit, tap)
• build.sh will perform a comparison of the expected test results (manually created by the author)