Combine near-duplicate test functions into parametrized tests

[jsh9]

This PR combines test functions that ran the same code with different inputs and expected results into parametrized tests (pytest.mark.parametrize).

Summary

In 15 of the 22 test files, 108 test functions become 39 parametrized tests with 109 cases. Every case still runs and is reported as its own test. Each test ID comes from the original test name or its comment, e.g. test_truncate[on_the_boundary_of_a_layer] or test_filter[band_pass].

The one extra case comes from test_loading_data__test_invalid_unit_names (ground motion). It had two checks, and they are now two cases of test_loading_data__invalid_input.

One more test is renamed, not combined: test_calc_damping_from_stress_strain__case_3 is now test_calc_damping_from_stress_strain__initial_damping_almost_0, after what it tests, because its two former siblings no longer have numbered IDs.

I compared the collected test IDs before and after: every removed test maps to exactly one new case (the renamed test maps to its new name), and the case count per file is the same (except the +1 above). All 241 tests pass, and pre-commit run -a passes.

What was combined

File	Combined tests
test_class_Vs_profile.py	12 format__case_* → test_Vs_profile_format; 5 truncate__case_* → test_truncate; 2 add_halfspace; f0_BH/f0_RO
test_class_curves.py	HH_x/H4_x fits (single layer, multi-layer DE, multi-layer DEAP); GGmax/damping get_curve_matrix
test_class_parameters.py	HH/MKZ multi-layer pairs (from file, from 2D array, file vs. array, list operations); 4 construct_curves; 2 param_serialize; 2 serialize_to_2D_array
test_class_ground_motion.py	3 PGA loading tests; the invalid-input tests; the 4 filters
test_class_site_factors.py	4 "not lenient" error tests; 6 "lenient" tests
test_class_svm.py	2 Vs_cap tests; 2 valid and 3 invalid get_discretized_profile tests
test_helper_site_response.py	calc_z1, calc_GGmax_from_stress_strain, calc_damping_from_stress_strain (cases 1–2), fit_all_damping_curves exceptions, align_two_time_arrays normal cases
test_helper_hh_calibration.py	calc_OCR, calc_K0, 4 optimization_kernel cases
test_helper_hh_model.py, test_helper_mkz_model.py	serialize/deserialize failure tests
test_class_batch_simulation.py	init cases 1–3
test_class_damping_calibration.py	get_HH_x_param/get_H4_x_param
test_class_hh_calibration.py	the 2 incorrect-Tmax tests
test_class_site_effect_adjustment.py	the 2 lenient tests
test_helper_simulations.py	elastic/rigid linear tests

Notes for review

• Method parameters: when cases call different methods of the same kind (the filters, get_HH_x_param()/get_H4_x_param(), get_f0_BH()/get_f0_RO()), the parameter is the unbound method (e.g. GM.highpass), not a method name passed to getattr.
• Keyword-only booleans: boolean parameters are keyword-only (*), following CONTRIBUTING.md, because muff reports FBT001 on boolean test arguments too.
• Behavior changes (small):
  • Expected-failure cases that used match='' now have no match. This is equivalent, and pytest no longer warns that an empty string always matches (13 warnings → 6).
  • The HH case of the multi-layer list-operations test checked len() twice. It now also checks n_layer, like the MKZ case.
  • The MKZ case of the multi-layer list-operations test now deletes layer 3, like the HH case, so that both cases can share one fixture (it used to delete layer 6). It still checks len() and n_layer afterwards.
  • The rigid case of test_linear now also labels the plot's y-axis, like the elastic case did.
• Not combined: tests whose assertions or setup differ, e.g. the HH vs. MKZ get_GGmax/get_damping tests, which build their objects with different classes and deserializers. Checks inside a single test function (e.g. test_is_int, test_range_check, test_query_Vs_given_thk) are also left alone, because this PR only combines separate test functions. Turning those into parametrized cases could be a follow-up.
• CHANGELOG.md has a new entry, and the example notebooks were re-run (second commit), as CONTRIBUTING.md requires.

🤖 Generated with Claude Code

https://claude.ai/code/session_01C1a5vwxyeL4atk9tYUUEoQ

---

Generated by Claude Code