Bug/fix ants initialization (when using only linear transforms)

[oesteban]

The generation of output filenames was incomplete for inverse outputs
in both collapsed and not-collapsed modes.

Additionally, collapsed mode for only-linear registrations was broken
as the warp file was not generated (but checked for existence)

[satra]

@oesteban - can you check why travis is failing?

[oesteban]

Sure, I looked at that, but did not find why it was failing:

FAIL: Doctest: nipype.interfaces.ants.registration.Registration
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/travis/miniconda/envs/testenv/lib/python2.6/doctest.py", line 2163, in runTest
    raise self.failureException(self.format_failure(new.getvalue()))
AssertionError: Failed doctest test for nipype.interfaces.ants.registration.Registration
  File "/home/travis/build/nipy/nipype/nipype/interfaces/ants/registration.py", line 326, in Registration
----------------------------------------------------------------------
File "/home/travis/build/nipy/nipype/nipype/interfaces/ants/registration.py", line 391, in nipype.interfaces.ants.registration.Registration
Failed example:
    reg5.cmdline
Expected:
    'antsRegistration --collapse-linear-transforms-to-fixed-image-header 0 --collapse-output-transforms 0 --dimensionality 3 --initial-moving-transform [ trans.mat, 1 ] --interpolation Linear --output [ output_, output_warped_image.nii.gz ] --transform Affine[ 2.0 ] --metric CC[ fixed1.nii, moving1.nii, 1, 4, Random, 0.05 ] --convergence [ 1500x200, 1e-08, 20 ] --smoothing-sigmas 1.0x0.0vox --shrink-factors 2x1 --use-estimate-learning-rate-once 1 --use-histogram-matching 1 --transform SyN[ 0.25, 3.0, 0.0 ] --metric CC[ fixed1.nii, moving1.nii, 0.5, 32, None, 0.05 ] --metric Mattes[ fixed1.nii, moving1.nii, 0.5, 32, None, 0.1 ] --convergence [ 100x50x30, 1e-09, 20 ] --smoothing-sigmas 2.0x1.0x0.0vox --shrink-factors 3x2x1 --use-estimate-learning-rate-once 1 --use-histogram-matching 1 --winsorize-image-intensities [ 0.0, 1.0 ]  --write-composite-transform 1'   
Got:
    'antsRegistration --collapse-linear-transforms-to-fixed-image-header 0 --collapse-output-transforms 0 --dimensionality 3 --initial-moving-transform [ trans.mat, 1 ] --interpolation Linear --output [ output_, output_warped_image.nii.gz ] --transform Affine[ 2.0 ] --metric CC[ fixed1.nii, moving1.nii, 1, 4, Random, 0.05 ] --convergence [ 1500x200, 1e-08, 20 ] --smoothing-sigmas 1.0x0.0vox --shrink-factors 2x1 --use-estimate-learning-rate-once 1 --use-histogram-matching 1 --transform SyN[ 0.25, 3.0, 0.0 ] --metric CC[ fixed1.nii, moving1.nii, 0.5, 32, None, 0.05 ] --metric Mattes[ fixed1.nii, moving1.nii, 0.5, 32, None, 0.1 ] --convergence [ 100x50x30, 1e-09, 20 ] --smoothing-sigmas 2.0x1.0x0.0vox --shrink-factors 3x2x1 --use-estimate-learning-rate-once 1 --use-histogram-matching 1 --winsorize-image-intensities [ 0.0, 1.0 ]  --write-composite-transform 1'

They are identical (or, at least, I don't see the difference). As I was to write a test for the new feature, I'll see if I'm making a mistake that crashes the cmdline()

[coveralls]

Coverage remained the same when pulling e1a9450 on oesteban:bug/FixAntsInitialization into 504858e on nipy:master.

[oesteban]

Ok, that was it :). But now travis fails because it can't install afni ¿?

[satra]

@oesteban - that error is because the neurodebian server does not respond from time to time, right @yarikoptic

@hjmjohnson - could you take a look at this PR? i think it looks good.

[satra]

@hjmjohnson - a quick peak would be appreciated :)

[oesteban]

I've found a case for which it fails, so please freeze this PR until I add the fix.

[coveralls]

Coverage remained the same when pulling f353ce4 on oesteban:bug/FixAntsInitialization into 504858e on nipy:master.

[coveralls]

Coverage remained the same when pulling eb972e4 on oesteban:bug/FixAntsInitialization into 504858e on nipy:master.

[oesteban]

I've checked that the PR now works OK.