replace .A and .A1 with explicit array conversion

[m-reuter]

Newer versions of SciPy seem to have forgotten the short functions .A() and .A1() for array conversion, so we call explicitly the long names.

[dkuegler]

@taha-abdullah please test this with scipy 1.14.1

[taha-abdullah]

I am getting the following error when I run this:

Traceback (most recent call last):
File "/fastsurfer/recon_surf/sample_parc.py", line 430, in
sample_parc(options.insurf, options.inseg, options.seglut, options.surflut, options.outaparc, options.incort, options.projmm, options.radius)
File "/fastsurfer/recon_surf/sample_parc.py", line 421, in sample_parc
smooths = smooth_aparc(surf, surfsamples, cortex)
File "/fastsurfer/recon_surf/smooth_aparc.py", line 335, in smooth_aparc
labels_new = mode_filter(adjM, labels, fillonlylabel, np.array([fillonlylabel]))
File "/fastsurfer/recon_surf/smooth_aparc.py", line 216, in mode_filter
sums = nlabels.sum(axis=1).toarray().squeeze()
AttributeError: 'matrix' object has no attribute 'toarray'
Command exited with non-zero status 1
@#@FSTIME 2024:08:23:11:00:19 recon_surf/sample_parc.py N 16 e 0.95 S 0.12 U 0.83 P 100% M 189276 F 0 R 51683 W 0 c 11 w 1 I 0 O 0 L 3.12 3.06 3.31
@#@FSLOADPOST 2024:08:23:11:00:20 recon_surf/sample_parc.py N 16 3.03 3.04 3.30

[dkuegler]

This error is with scipy 1.14.1 u
I assume?

[m-reuter]

I assumed that sum produces a sparse matrix just as max, but it produces a matrix, so array conversion is slightly different. I also fixed deprecated in1d. These changes have now been tested with numpy 2.1.0 and script 1.14.1 and yield identical results to the previous version in my test case.