Better output for multiple (failing) tests with --as-cran

[gaborcsardi]

Closes #161.

@hadley

OK, now the you get this with --as-cran:

─  checking examples ... NONE (505ms)
✔  checking for unstated dependencies in ‘tests’ ...
─  checking tests (490ms)
─  Running ‘a.R’ (523ms)
─  Running ‘b.R’ (491ms)
─  Running ‘c.R’ (532ms)
E  Some test files failed
   Running the tests in ‘tests/b.R’ failed.
   Complete output:
     > stop("error")
     Error: error
     Execution halted
✔  checking PDF version of manual (6.1s)

and this without --as-cran:

✔  checking for unstated dependencies in ‘tests’ ...
─  checking tests (454ms)
✔  Running ‘a.R’ (508ms)
E  Running ‘b.R’ (503ms)
   Running the tests in ‘tests/b.R’ failed.
   Complete output:
     > stop("error")
     Error: error
     Execution halted
✔  checking PDF version of manual (6.9s)

Ideally we would look at the .Rout and .Rout.fail files, and mark the failed test files in the --as-cran case as well, but that would need a bigger rewrite, because right now the output is based on the stdout + stderr of the check process only.

[hadley]

Thanks for the fix!