Coverage for failing tests

[briederer]

Adds a safety measure to still process .cov files even when tests are failing.

Closes #57

[briederer]

Sorry for messing with the tests. I wanted to achieve that the tests for this package are succeeding while in general they generate_coverage still fails. Therefore it was neccessary to add the should_throw argument to test_coverage function. This checks now that the .cov files are processed correctly even if generate_coverage still fails when one or more tests are failing.

[briederer]

@tpapp Ready for review

[briederer]

Thanks for the review. Added a clarification comment to your question

[briederer]

Short reminder

[tpapp]

Thanks, I appreciate the reminder. The code now looks good to me.

I am wondering if proceeding in case of an error is always desirable. Would an argument that controls this for generate_coverage make sense? The new behavior could be the default.

Can you please add a section (eg # Error handling) to the documentation of generate_coverage that explains the new behavior? Just that errors are caught and coverage is still generated.