Reduce Warnings

[Scienfitz]

This is mostly a housekeeping PR aimed at reducing unnecessary warnings. Examples are for instance unnecessary tensor or integer conversion, pandas behavior that is deprecated etc.

Two sources of warnings have not been touched:

• Use of Dummy in discrete constraints: FutureWarning from pandas but likely more complex to fix
• test_deprecations uses construction of objects in parameterization that is deprecated. Since this happens before code execution warnings filters did not have any effect on this (including a file wide pytestmark)

A full run of all tests with pytest -vv leads to 246 warnings on a fairly up to date dev environment versus 722 warnings previously.

[Copilot]

Pull request overview

Housekeeping PR focused on reducing noisy/deprecated warnings across the codebase (tests, docs, and a couple of runtime utilities), mainly around pandas and torch behaviors.

Changes:

• Suppress or avoid warnings in tests (e.g., UnusedObjectWarning, pandas dtype FutureWarnings, SHAP runtime warnings).
• Adjust dataframe dtype normalization to avoid warning on integer-to-float conversions while still converting internally.
• Avoid pandas FutureWarnings in constraint logic and docs examples (copying slices; avoiding concat with empty frames).

Reviewed changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
tests/validation/test_campaign_validation.py	Suppresses UnusedObjectWarning during overlapping-name validation for non-predictive recommenders.
tests/utils/test_sampling_algorithms.py	Conditionally suppresses the “All points are identical” warning in FPS property-based tests.
tests/utils/test_dataframe.py	Adds an integer-valued “valid” case for dtype normalization tests.
tests/test_simulation.py	Adds include_groups=False to silence pandas groupby/apply warnings.
tests/test_settings.py	Suppresses UnusedObjectWarning during caching tests for non-predictive recommenders.
tests/test_kernels.py	Replaces torch.tensor(...) with torch.as_tensor(...) to reduce tensor-construction warnings.
tests/test_input_output.py	Uses .loc assignment and suppresses pandas FutureWarning for deliberate bad-dtype injections.
tests/serialization/test_json_serialization.py	Ensures opened file handles are closed even when JSON decoding raises.
tests/insights/test_shap.py	Suppresses SHAP-internal RuntimeWarning noise while still asserting expected UserWarnings.
docs/userguide/async.md	Uses .copy() when slicing to avoid pandas chained-assignment/view warnings.
baybe/utils/dataframe.py	Refines dtype normalization warning logic (don’t warn for integer columns, still convert).
baybe/constraints/discrete.py	Avoids pandas FutureWarning by skipping concat with empty DataFrame parts.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[AdrianSosic]

Thx!