test: Add tests for quality-control functionality

[madhavmk]

Currently quality control functionality is not covered with pytest.

[madhavmk]

Un-assigning myself from this Issue. This is a great first issue to solve to gain an understanding of the solver module.

[weknowthecalmat]

Hi I am Vinod Kamat, a graduate student at University of Washington. I wanted to work on this issue. A few key observations on this -

I'm interested in contributing to GNATSS and would like to work on the quality control testing issue. After looking at the codebase, I've observed that while the core quality control functions in src/gnatss/ops/validate.py (like calc_std_and_verify(), check_sig3d(), and check_solutions()) are implemented and working correctly when tested manually, they currently lack comprehensive pytest coverage. The existing test suite in tests/ops/test_qc.py only covers plotting utilities (_compute_ticks_and_labels), while critical scientific validation functions that enforce GPS accuracy limits, filter outliers, and validate mathematical convergence remain untested. When I was going through the codebase and running the application, I saw that these QC functions work as expected - for example, calc_std_and_verify() correctly rejects GPS data exceeding the 5cm accuracy threshold, and check_sig3d() properly filters out bad positioning data. However, without automated tests, changes to these functions could introduce bugs that compromise the scientific integrity of seafloor positioning results. I'd like to create comprehensive pytest coverage for these validation functions, including edge cases, error conditions, and parameter validation. Could you assign this issue to me? Also, are there any specific testing patterns or scientific accuracy requirements I should follow when writing these tests?

[atambay37]

Hi @weknowthecalmat, bringing in the maintainer here: @johnbdesanto. John, could you see if Vinod is on the right track here and assign him the issue if so?

[weknowthecalmat]

Thanks @atambay37 !

[johnbdesanto]

Hi @weknowthecalmat, @atambay37. I have glanced through the testing infrastructure and it does appear that we don't have checks set up for the residual flagging, just the residual calculation (although I suspect this test is skipped due to incompatibilities between numba and pytest).

Since the subroutines you identified are not numba code I see no problems with creating tests for them. There are probably other functions that could use more robust testing infrastucture as well. I'll go ahead and assign this to you, thanks.

[weknowthecalmat]

Thanks @johnbdesanto , I will work on this.