Remove remaining scda usage, update to testthat v3.0

[edelarua]

Closes #709

[github-actions]

Unit Tests Summary

    1 files    33 suites   3s ⏱️
149 tests 149 ✔️     0 💤 0 ❌
280 runs  168 ✔️ 112 💤 0 ❌

Results for commit c85816d.

♻️ This comment has been updated with latest results.

[github-actions]

Unit Test Performance Difference

Additional test case details

Test Suite	$Status$	Time on main	$±Time$	Test Case
tm_t_shift_by_grade	💀	$0.43$	$-0.43$	template_shift_by_grade_is_keeping_the_same_number_of_missing_data_as_Missing_at_the_end_of_preprocessing_steps
tm_t_shift_by_grade	👶		$+0.20$	template_shift_by_grade_keeps_the_same_number_of_missing_data_Missing_after_preprocessing
tm_t_shift_by_grade	👶		$+0.02$	template_shift_by_grade_throws_an_error_when_worst_flag_var_is_not_one_of_WGRLOVFL_WGRLOFL_WGRHIVFL_WGRHIFL
tm_t_shift_by_grade	💀	$0.01$	$-0.01$	template_shift_by_grade_throws_an_error_when_worst_flag_var_is_not_one_of_WGRLOVFL_WGRLOFL_WGRHIVFL_WGRHIFL_
tm_t_summary	👶		$+0.02$	template_summary_generates_correct_expressions_for_multiple_grouping_variables_and_all_patients
tm_t_summary	💀	$0.01$	$-0.01$	template_summary_generates_correct_expressions_for_multiple_grouping_variables_and_all_patientts

Results for commit 4053160

♻️ This comment has been updated with latest results.

[shajoezhu]

thanks a lot @edelarua , looks really good. I will take another closer look later.

[nikolas-burkoff]

A few comments from me

[Melkiades]

Thank you Emily for all the work! I am only a bit unsure if changing everything into e3 is in scope for this issue and PR. Also, it seems to me that a lot of the expected is built here around alternative ways to produce the same result which is valuable information. Changing it to e3 makes it also difficult to read the scda-related changes

[Melkiades]

After our discussion, I reckon that the snapshots and the expected evaluations are both text-based and, therefore, using the first in place of the second makes sense. The code is now much cleaner and the errors/changes will be more readable in the future. It is still a large PR (due to the shift of info from tests to snaps) and it should have been done in a separate PR. Nonetheless, the modifications related to scda were minimal and, therefore the vbump problem is minor. Surely, we should (myself too) try to remember that it is better to do one issue per PR so that vbump can be consulted better. Overall, GREAT job Emily!

[edelarua]

@shajoezhu are we good to merge this or do you need time to look it over?

[shajoezhu]

looks great! Thanks so much! @edelarua