Allow blessing if test only failed GENERATE

[samsrabin]

Sometimes a test might fail only in the GENERATE step; for example, if the user doesn't have write permissions in the baseline directory. This PR makes it so that bless_test_results does try to bless things in such cases.

Test suite:
Test baseline:
Test namelist changes:
Test status: [bit for bit, roundoff, climate changing]

Fixes: None

User interface changes?: No

Update gh-pages html (Y/N)?: No

[jedwards4b]

I would like to have a review from one of the e3sm developers to make sure that this doesn't adversely affect them.

[samsrabin]

The one failing unit test is test_is_bless_needed_overall_fail. Seems like its name and assertions should be changed to reflect that a test failing only in GENERATE should pass blessing. Then another test should be added to check for overall failure due to something other than just GENERATE.

[jasonb5]

@jgfouca What do you think about this functionality for E3SM?

[jgfouca]

I think this change is fine for E3SM.

[jedwards4b]

@samsrabin does the unit-testing need to be fixed?

[jasonb5]

Just need testing to pass and I'll review.

[samsrabin]

I'm having trouble fixing the unit testing because I can't figure out where the SMS.f19_g16.A test is being set up. Could one of y'all point me to that?

[jasonb5]

@samsrabin The test is setup by mocking TestStatus object which is passed to is_hist_bless_needed.

I would update ts.get_status.side_effect = ["PASS", "FAIL"] + ["PASS"] * (len(ALL_PHASES) - 1) to ts.get_status.side_effect = ["PASS",]*len(ALL_PHASES)+1 so the function will still test the original intended behavior.

Then can copy it and update it with ts.get_status.side_effect = ["PASS",]*10 + ["FAIL",] + ["PASS",]*5. This should test when the GENERATE_PHASE is marked as FAIL.