ENH: add test for FBTractMeasure with scalar-only polydata

[ihnorton]

Adds test for FiberTractMeasurements CLI module with scalar-only polydata.

[ihnorton]

Data is a single-strand fiberbundle from UKF.

Added random scalar array in Python (assuming pd points to the polydata already):

In [75]: randarr = np.random.rand(60)

In [76]: vrandarr = dsa.numpyTovtkDataArray(randarr)

In [77]: vrandarr
Out[77]: (vtkDoubleArray)0x110ddc350

In [78]: np.mean(randarr)
Out[78]: 0.57747233392019237

In [79]: vrandarr.SetName("RANDARRAY")

In [80]: pd.GetPointData().AddArray(vrandarr)
Out[80]: 1

In [81]: wr = vtk.vtkPolyDataWriter()

In [82]: wr.SetFileName("/tmp/fbt/littlerandbundle.vtk")

In [83]: wr.SetInputData(pd)

In [84]: wr.Update()

In [85]: wr.Write()

[ihnorton]

Manual verification of figures in baseline output file:

# RANDARRAY values from `fiber_notensor_randscalar.vtk`
julia> v = [0.81787 0.595397 0.197063 0.991127 0.976741 0.804367 0.346301 0.100142 0.193635 0.0378677 0.444774 0.780073 0.426525 0.774192 0.765277 0.968237 0.069852 0.287893 0.0604638 0.340746 0.98026 0.611797 0.963014 0.878509 0.91017 0.721776 0.0652108 0.789717 0.639065 0.673024 0.0148516 0.923146 0.388264 0.920742 0.485409 0.99897 0.414107 0.562507 0.474771 0.182289 0.621647 0.737416 0.96598 0.627021 0.463968 0.928957 0.518274 0.79058 0.99335 0.396943 0.27392 0.0251089 0.995054 0.46926 0.0960025 0.301948 0.319466 0.878315 0.772408 0.896582 ]
julia> minimum(v)
0.0148516

julia> maximum(v)
0.99897

julia> mean(v)
0.5774723883333334

julia> median(v)
0.616722

julia> minimum(v)
0.0148516

julia> var(v)
0.09934554264379902

vs

Max	Mean	Median	Min	NAN	Variance
0.998970	0.577472	0.621647	0.014852	0.000000	0.099346

[ihnorton]

After putting the numbers in the table above, I noticed variance median was incorrect. This was because the calculation did not take in to account the possibility of even numbers of points. Proposed fix added to this PR in eed6624.

Corrected table after fix:

Max	Mean	Median	Min	NAN	Variance
0.998970	0.577472	0.616722	0.014852	0.000000	0.099346

[ihnorton]

I noticed variance median was incorrect. This was because the calculation did not take in to account the possibility of even numbers of points. Proposed fix added to this PR in eed6624.

cc @zhangfanmark since you wrote this code (I think), can you take a look at this change?

[zhangfanmark]

Hi @ihnorton, thanks for the update. The changes look good to me.