Add dataset-level attribute comparison support (#148)

[botsp]

Summary

This PR implements #148 by adding dataset-level attribute comparison in diffdf(), so differences such as dataset label (e.g., DM = "Demographics") can be detected and reported.

Closes #148.

What changed

• Added a new internal helper in R/identify.R:
  • identify_df_att_differences(BASE, COMP, exclude_attrs = c("names", "row.names", "class"))
  • Compares dataframe-level attributes between BASE and COMPARE.
• Updated R/diffdf.R:
  • Captures dataframe-level attribute differences early in the workflow.
  • Adds a new issue component:
    • DataframeAttribDiffs
    • message: BASE and COMPARE dataframes have differing attributes !!
• Updated tests in tests/testthat/test-core.R:
  • Added checks that dataframe-level label differences trigger warnings.
  • Added checks that DataframeAttribDiffs has expected row count.

Validation

Unit test (local source)

• testthat::test_local('.', filter = 'core')
• Result: PASS (102)

Real XPT checks (3 scenarios)

Using SDTM XPT examples, validated:

1. Fully consistent datasets
   • Result: No issues were found!
2. Dataset label inconsistent
   • Result: DataframeAttribDiffs reports label difference.
3. Dataset label + variable label inconsistent
   • Result: Both dataframe-level and variable-level attribute differences are reported in print output.

Notes

• Existing variable-level attribute comparison (AttribDiffs) behavior is preserved.
• This PR only adds dataset-level attribute detection/reporting and related tests.

[gowerc]

Hey @botsp ,

Thanks for this looks good, only remark from me would be could we just have a print test for more complex attributes to make sure it doesn't break the print method. I mean the code is basically the same construct as what is already used so I suspect its fine but would be good to just assert so with a test.

[botsp]

hi @gowerc Thanks for the review. Yes -- for #148 I intentionally followed the existing code structure used in the package. And thanks for the kind reminder, I also added print regression tests for dataframe-level attributes (including complex nested attributes, NULL vs non-NULL, character vector attributes, and a Japanese dataset label case) to confirm the print path remains stable.
Could you please take another look when you have a chance?