Handle some TODOs in the model workflow

[mwaskom]

A few changes here:

• Fixed smoothing in the model workflow. Previous inplace smoothing was silently failing because the images were loaded as np.float32 and then doing astype(np.float) was forcing a copy. Some enhancements / more testing in the signals module and new code in the model workflow address this.
• Fixed (provisionally) the definition of cortical voxels. It turns out that the surface vertex image does not always agree with the nearest-neighbor-resampled wmparc image (generating seg.nii.gz), so some voxels with an identified cortical vertex were being considered white matter/out of brain/etc. This needs to be handled upstream, however.
• The logic of not smoothing / filtering is handled within the underlying library functions so that there is less conditional complexity in the workflow code itself.
• Made noise interpolation default to False until it can be properly parameterized.

[codecov-io]

Codecov Report

Merging #150 into master will not change coverage. The diff coverage is 100%.

@@          Coverage Diff          @@
##           master   #150   +/-   ##
=====================================
  Coverage     100%   100%           
=====================================
  Files          10     10           
  Lines        2046   2077   +31     
=====================================
+ Hits         2046   2077   +31

Impacted Files	Coverage Δ
lyman/surface.py	100% <ø> (ø)	:arrow_up:
lyman/frontend.py	100% <ø> (ø)	:arrow_up:
lyman/signals.py	100% <100%> (ø)	:arrow_up:
lyman/glm.py	100% <100%> (ø)	:arrow_up:
lyman/workflows/model.py	100% <100%> (ø)	:arrow_up:
lyman/__init__.py	100% <100%> (ø)	:arrow_up:

---

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 6872237...433b662. Read the comment docs.