Review slow tests

[dagewa]

Now all tests are using dials-data we thought it might be a good time to review test durations. Here are the top twenty slowest DIALS tests:

206.40s call tests/algorithms/integration/test_simple_integrate.py::test_against_dials_integrate
117.40s call tests/algorithms/scaling/test_tof_scaling.py::test_tof_extract_shoeboxes

100.79s call     tests/algorithms/symmetry/cosym/test_cosym_analyse_datasets.py::test_cosym[I23-None-dimensions2-10-False-False-True]

91.56s call tests/command_line/test_model_background.py::test_model_background
89.77s call tests/algorithms/indexing/test_index.py::test_pink_indexer

55.71s call     tests/algorithms/indexing/test_index.py::test_index_small_molecule_multi_sequence_3
54.57s call     tests/command_line/test_integrate.py::test_multi_lattice
45.05s call     tests/command_line/test_ssx_reduction.py::test_ssx_reduction
42.57s call     tests/test_scan_varying_integration_bug.py::test
40.57s call     tests/algorithms/indexing/test_index.py::test_unconventional_P1_cell[cell_params1]
40.16s call     tests/algorithms/profile_model/ellipsoid/test_refiner.py::test_Refiner
39.26s call     tests/algorithms/indexing/test_index.py::test_unconventional_P1_cell[cell_params0]
38.81s call     tests/command_line/test_preservation_of_experiment_identifiers.py::test_for_preservation_of_identifiers_in_dials_processing
35.93s call     tests/algorithms/indexing/test_index.py::test_index_multi_lattice_multi_sweep
31.20s call     tests/command_line/test_refine.py::test_scan_varying_refinement_rmsds
30.10s call     tests/algorithms/indexing/test_index.py::test_index_4rotation
29.48s call     tests/command_line/test_spot_counts_per_image.py::test_spot_counts_per_image[thaumatin_grid_scan]
29.27s call     tests/command_line/test_detect_blanks.py::test_passing_still_images_raises_sysexit
28.71s call     tests/command_line/test_ssx_index.py::test_ssx_index_input_unit_cell
28.56s call     tests/algorithms/scaling/test_scale.py::test_incremental_scale_workflow

We might pick on one or two of these at a DIALS catch up in the near future.

[dagewa]

As of today the slowest 20 DIALS tests determined by serial execution via pytest --regression --durations=20 are:

36.11s call     tests/command_line/test_ssx_reduction.py::test_ssx_reduction
27.76s call     tests/algorithms/profile_model/ellipsoid/test_refiner.py::test_Refiner
27.73s call     tests/command_line/test_integrate.py::test_multi_lattice
25.91s call     tests/command_line/test_model_background.py::test_model_background
25.41s call     tests/command_line/test_preservation_of_experiment_identifiers.py::test_for_preservation_of_identifiers_in_dials_processing
23.07s call     tests/algorithms/indexing/test_index.py::test_index_multi_lattice_multi_sweep
20.57s call     tests/algorithms/indexing/test_index.py::test_index_4rotation
19.81s call     tests/command_line/test_refine.py::test_scan_varying_refinement_rmsds
19.56s call     tests/algorithms/symmetry/test_laue_group.py::test_determine_space_group_best_monoclinic_beta
19.12s call     tests/algorithms/scaling/test_scale.py::test_incremental_scale_workflow
19.02s call     tests/command_line/test_ssx_index.py::test_ssx_index_input_unit_cell
18.67s call     tests/command_line/test_search_beam_position.py::test_index_after_search
16.49s call     tests/command_line/test_stills_process.py::test_sacla_h5[control_flags4]
16.18s call     tests/command_line/test_spot_counts_per_image.py::test_spot_counts_per_image[thaumatin_grid_scan]
15.48s call     tests/algorithms/indexing/test_index.py::test_pink_indexer
15.13s call     tests/algorithms/integration/test_filter_overlaps.py::test_for_overlaps
14.36s call     tests/algorithms/indexing/test_index.py::test_all_expt_ids_have_expts
13.83s call     tests/util/test_image_grouping.py::test_real_cbf_example
13.38s call     tests/algorithms/indexing/test_index.py::test_refinement_failure_on_max_lattices_a15
13.19s call     tests/command_line/test_integrate.py::test_imageset_id_output_with_multi_sweep

Quite a bit better! I don't know what happened to test_cosym_analyse_datasets.py. Maybe last time I had run with -n auto and one of those tests started but got stalled? Anyway, the new ranking here should be trustworthy, and no longer looks so bad.