Fail loudly if new benchmark has no output.

[n8xm]

When --make_benchmarks is on, a misconfigured / buggy test program might fail to produce any output. If this happens, the regression testing script will continue executing and eventually crash when it can't find the .status file that it is expecting to be there, creating an unintuitive backtrace that does not make clear the root cause of the problem:

Traceback (most recent call last):
  File "./regtest.py", line 1317, in <module>
    n = test_suite(sys.argv[1:]) File "./regtest.py", line 1245, in test_suite
    test_list, args.input_file[0])
  File "/home/nathan/Code/regression_testing/test_report.py", line 954, in report_this_test_run with 
    open(benchStatusFile) as bf:
FileNotFoundError: [Errno 2] No such file or directory: 'TestName.status' 

This PR fixes the above by failing with a clear error message as soon as it can be determined that output from the test program cannot be found.

[ax3l]

Oh no, this broke WarpX #136.

Can we make this optional?

[zingale]

I'm okay with making it optional -- @n8xm are you okay with making this off by default and enabled with a runtime parameter?

[zingale]

this breaks our test suites

I am going to make it optional