Improve tests

[JorMaister]

• Goes with https://github.com/scipion-em/scipion-em-tomo/pull/470
• The tests now use the Test Centralization Layer update and expansion and the used dataset expanded definition.
• Tests passing

[azazellochg]

Hi! I assume you want to merge to devel not master?

[JorMaister]

Hi, @azazellochg,

As it's a test update, it may remain in devel until more changes are sent to devel before making a release.

[azazellochg]

I'm confused, this PR targets the master, which contradicts with what you are saying

[JorMaister]

My fault, @azazellochg! I didn't notice that. I'll change it now.

Many thanks

[JorMaister]

Hi, @azazellochg, They're passing for me. Could you confirm If you are in the latest state of scipion-em-tomo (devel)?

[azazellochg]

Hi, yes I can confirm tomo in devel. I'm using: ARETOMO_HOME = /public/EM/Scipion/scipion-dev/software/em/aretomo2-1.0.0 ARETOMO_CUDA_LIB = /public/EM/CUDA/cuda-11.8/lib64 ARETOMO_BIN = AreTomo2_1.0.0_Cuda118

[JorMaister]

May it indicate that the refined rotation angle is different depending on the binary used? Does it fail using aretomo branch devel branch? I'm using cuda 12.2... But the same before the changes... Let me go back to devel and re-execute the tests

[JorMaister]

Test passing after having moved to devel in aretomo... Does it happen to you, @azazellochg ?

[azazellochg]

Hi, I think your guess is correct. It may be the binary or the GPU that gives different results. I get the same error in devel branch. Maybe we can relax the tolerance a bit?

[JorMaister]

This case is very specific... I think some tolerance thresholds could be added as optional inputs in the checking methods, leaving the ones that currently exist as default and leaving the option of modifying them for specific cases, such as this one. Thanks, @azazellochg