add SAS codestub and autograde for metakernel based non-python kernels

[jld23]

similar to #870 I've added a code stub to get people started using nbgrader for teaching SAS.
This is a one line change.

The more consequential change is supporting auto-grading when cells return as a stream and then output.name is stderr.
This issue was discussed in #1173 specifically for octave and @jhamrick indicated support for a change to compare output.output_type == "stream" and output.name == "stderr":

This change is useful because it allows the many non-Python language kernels based on metakernel, not just octave, to integrate with nbgrader if they use the metakernel Error or Error_display methods.

[jld23]

I've never used codecov before it appears the check is failing on my typo commit. I'm not sure how to remediate.

[jld23]

Is there anything needed from me to advance this PR?

[BertR]

Hi @jld23, thanks for this! The change looks good to me. I can't figure out why the Windows build is failing, but it might not have anything to do with your PR. Would you be able to check your code on a Windows environment?

[jld23]

@BertR, I got a windows system working and I can confirm I'm seeing issues with 0.7.0.dev0 (my PR branch) and not with pip installed 0.6.1 I'm getting errors where I don't think my code is involved like updating the notebooks nbgrader update source.

Not sure how you want to proceed.

[BertR]

@BertR, I got a windows system working and I can confirm I'm seeing issues with 0.7.0.dev0 (my PR branch) and not with pip installed 0.6.1 I'm getting errors where I don't think my code is involved like updating the notebooks nbgrader update source.

Not sure how you want to proceed.

@jld23 it's possible this failed due to broken pypi dependencies in the master branch which have now been fixed.
Could you rebase and push again?