FIX: Build filter_basis of correct size

[effigies]

As currently coded, an empty filter basis that then has non-steady-state outlier regressors added in does not have the correct size. It is supposed to have the same number of rows as the confound file, but instead is a square matrix (same number of rows as outliers).

An empty filter basis is possible for short BOLD series, which I suspect will only come up in tests, but it is coming up in our tests.

Changes proposed in this pull request

• Creates final filter basis array with same number of rows as the confounds array

Follow-up to #2107, #2122, #2136.

[codecov-io]

Codecov Report

Merging #2138 into master will decrease coverage by 0.86%.
The diff coverage is 0%.

@@            Coverage Diff             @@
##           master    #2138      +/-   ##
==========================================
- Coverage   72.18%   71.32%   -0.87%     
==========================================
  Files        1158     1154       -4     
  Lines       58020    57719     -301     
  Branches     8334     8328       -6     
==========================================
- Hits        41884    41166     -718     
- Misses      14814    15173     +359     
- Partials     1322     1380      +58

Flag	Coverage Δ
#smoketests	71.32% <0%> (-0.87%)	⬇️
#unittests	69.82% <0%> (-0.02%)	⬇️

Impacted Files	Coverage Δ
nipype/algorithms/confounds.py	67.31% <0%> (ø)	⬆️
nipype/interfaces/fsl/model.py	54.79% <0%> (-25.57%)	⬇️
nipype/workflows/fmri/fsl/preprocess.py	73.35% <0%> (-14.62%)	⬇️
nipype/workflows/fmri/fsl/estimate.py	47.88% <0%> (-11.27%)	⬇️
nipype/interfaces/fsl/preprocess.py	74.96% <0%> (-7.64%)	⬇️
nipype/interfaces/fsl/utils.py	68.09% <0%> (-4.64%)	⬇️
nipype/algorithms/modelgen.py	70.93% <0%> (-2.56%)	⬇️
nipype/interfaces/fsl/base.py	81.91% <0%> (-2.13%)	⬇️
nipype/utils/filemanip.py	83.22% <0%> (-1.68%)	⬇️
nipype/interfaces/fsl/maths.py	98.26% <0%> (-1.16%)	⬇️
... and 9 more

---

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 d9b183b...9be01f2. Read the comment docs.