[FIX] CircleCI tests - error codes are not propagated

[oesteban]

The former structure of the tests.sh file in which all the tests are called
did not exit with error if one of the test commands errored, since the
last exit code was from codecov, and no exit code tracking or propagation
was done.

This PR fixes that issue. Additionally, the scripts related to CircleCI
have been split up into a new .circle/ folder.

[oesteban]

@chrisfilo I have some concerns with the collection of coverage results. Looking at the artifacts, it seems like only the node 0 is generating coverage xmls from the workflows.

My question is: are the rest of nodes not supposed to generate coverage results?

Other than that (and it could be fixed in another PR since we probably want this to be merged ASAP), this is ready to merge in.

[codecov-io]

Codecov Report

Merging #1858 into master will increase coverage by 0.07%.
The diff coverage is 75%.

@@            Coverage Diff             @@
##           master    #1858      +/-   ##
==========================================
+ Coverage   73.04%   73.11%   +0.07%     
==========================================
  Files        1064     1064              
  Lines       53366    53370       +4     
==========================================
+ Hits        38980    39023      +43     
+ Misses      14386    14347      -39

Flag	Coverage Δ
#unittests	73.11% <75%> (+0.07%)	✅

Impacted Files	Coverage Δ
...ype/interfaces/freesurfer/tests/test_preprocess.py	96.22% <75%> (+4.38%)	✅
nipype/utils/filemanip.py	87.17% <0%> (+0.36%)	✅
nipype/interfaces/tests/test_runtime_profiler.py	44.13% <0%> (+0.55%)	✅
nipype/interfaces/utility/wrappers.py	89.79% <0%> (+1.02%)	✅
nipype/pipeline/plugins/multiproc.py	77.09% <0%> (+1.67%)	✅
nipype/utils/tests/test_cmd.py	98.55% <0%> (+2.89%)	✅
nipype/init.py	68.75% <0%> (+3.12%)	✅
nipype/utils/tests/test_filemanip.py	93.21% <0%> (+12.66%)	✅

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e450340...26dfd79. Read the comment docs.

[codecov-io]

Codecov Report

Merging #1858 into master will increase coverage by 0.07%.
The diff coverage is 75%.

@@            Coverage Diff             @@
##           master    #1858      +/-   ##
==========================================
+ Coverage   73.04%   73.11%   +0.07%     
==========================================
  Files        1064     1064              
  Lines       53366    53370       +4     
==========================================
+ Hits        38980    39023      +43     
+ Misses      14386    14347      -39

Flag	Coverage Δ
#unittests	73.11% <75%> (+0.07%)	✅

Impacted Files	Coverage Δ
...ype/interfaces/freesurfer/tests/test_preprocess.py	96.22% <75%> (+4.38%)	✅
nipype/utils/filemanip.py	87.17% <0%> (+0.36%)	✅
nipype/interfaces/tests/test_runtime_profiler.py	44.13% <0%> (+0.55%)	✅
nipype/interfaces/utility/wrappers.py	89.79% <0%> (+1.02%)	✅
nipype/pipeline/plugins/multiproc.py	77.09% <0%> (+1.67%)	✅
nipype/utils/tests/test_cmd.py	98.55% <0%> (+2.89%)	✅
nipype/init.py	68.75% <0%> (+3.12%)	✅
nipype/utils/tests/test_filemanip.py	93.21% <0%> (+12.66%)	✅

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e450340...26dfd79. Read the comment docs.

[chrisgorgo]

Since unittests are only run on node 0 it is the only one generating coverage XML files. Other nodes run smoke tests and generate other XML files (which from what I can see are also uploaded).

[oesteban]

@chrisfilo shouldn't this line at the end of the build output of node 2 generate an xml file?:

coverage xml -o /scratch/smoketest_fmri_spm_nested_MultiProc__root_examples__l2pipeline.xml

Apparently that is not happening now...