feat: Add example rows to ValidationError for all rule failures

[Copilot]

Validation failures only reported a row count, requiring a debugger session to identify which values caused failures—painful in long-running pipelines.

Changes

• src/polars_plugin/validation_error.rs: Extended format_rule_failures to accept an optional examples parameter (dict[str, list[str]]). Updated RuleValidationError::to_string to include example rows for both schema-level and column-level rules.
• src/polars_plugin/mod.rs: Updated all_rules_required polars plugin to accept data columns as additional inputs after the rule boolean columns. For each failing rule, computes up to 5 distinct example rows using AnyValue::Display and includes them in the lazy-execution error message.
• dataframely/filter_result.py: Added public FailureInfo.examples(max_examples=5) helper method that returns distinct example rows (as formatted strings) for each failing rule.
• dataframely/_plugin.py: Added data_columns parameter to all_rules_required; passes data columns as additional args alongside num_rule_columns kwarg.
• dataframely/schema.py: Eager path calls failure.examples() and passes results to format_rule_failures. Lazy path passes cls.column_names() as data_columns to all_rules_required.
• dataframely/collection/collection.py: Updated format_rule_failures call to pass failure.examples().
• dataframely/_native.pyi: Updated type stub for format_rule_failures.
• tests/schema/test_validate.py: Updated test_invalid_primary_key to assert examples appear in the error message for both eager and lazy paths.

Example

# Before
dataframely.exc.ValidationError: 1 rules failed validation:
 - 'primary_key' failed for 312 rows

# After (eager)
dataframely.exc.ValidationError: 1 rules failed validation:
 - 'primary_key' failed for 312 rows with 4 distinct examples: [{'user_id': 'abc', 'name': 'Alice'}, ...]

# After (column rule, eager)
dataframely.exc.ValidationError: 1 rules failed validation:
 * Column 'name' failed validation for 1 rules:
   - 'max_length' failed for 5 rows with 3 distinct examples: [{'user_id': 'x', 'name': 'toolongvalue'}, ...]

Examples are also included in lazy execution errors raised by the Rust plugin. Note that string values use double quotes in lazy errors (Rust AnyValue::Display format) versus single quotes in eager errors (Python str(dict) format).

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[borchero]

• Adjust the format_rule_failures method to optionally include examples. Possibly add a helper method to FailureInfo if that is helpful.
• Do not only provide examples for primary key violations but for any type of validation
• Adjust the Rust plugin s.t. the same information is generated during lazy execution

[Copilot]

• Adjust the format_rule_failures method to optionally include examples. Possibly add a helper method to FailureInfo if that is helpful.
• Do not only provide examples for primary key violations b...

Addressed in commit 462f8ce:

1. format_rule_failures extended: Now accepts an optional examples: dict[str, list[str]] | None parameter. Each value is a list of pre-formatted example row strings. The examples are shown for both schema-level and column-level rules.

2. Examples for all rule types: Added FailureInfo.examples() helper method (the suggested helper on FailureInfo) that computes distinct example rows from all data columns for every failing rule. This is used in both schema.py and collection/collection.py eager validation paths.

3. Lazy execution (Rust plugin): Updated all_rules_required to accept data columns as additional inputs after the rule boolean columns (num_rule_columns kwarg tells it where the split is). For each failing rule, the plugin computes up to 5 distinct example rows using AnyValue::Display. The schema.py lazy path now passes cls.column_names() as data columns.

Example output:

# Eager
1 rules failed validation:
 - 'primary_key' failed for 2 rows with 2 distinct examples: [{'a': 1, 'b': 'x', 'c': '1'}, {'a': 1, 'b': 'y', 'c': '2'}]

# Lazy
1 rules failed validation for schema 'MySchema':
 - 'primary_key' failed for 2 rows with 2 distinct examples: [{'a': 1, 'b': "x", 'c': "1"}, {'a': 1, 'b': "y", 'c': "2"}]

Note: string quotes differ slightly between eager (Python str(dict) → single quotes) and lazy (Rust AnyValue::Display → double quotes), but both are readable.