Extend expression planners for Snowflake compatibility

[osipovartem]

Summary

• add backward-compatible schema-aware hooks for aggregate and scalar expression planners
• preserve the existing planner methods as defaults, so upstream behavior is unchanged
• pass the function input schema from SqlToRel
• plan qualified wildcard arguments and wildcard options such as alias.* EXCLUDE (...)
• add an opt-in datafusion.sql_parser.trim_decimal_literal_trailing_zeros setting for Snowflake-compatible exact decimal inference
• update h2 to 0.4.16 to address RUSTSEC-2026-0258

This is the collapsed DataFusion side of the Rustice Snowflake compatibility stack. It includes the work previously reviewed independently in #48, #49, and #50.

Compatibility

All new behavior is opt-in through custom expression planners or the new configuration setting. Default DataFusion parsing, type inference, and expression planning remain unchanged.

The hooks allow Rustice to implement schema-dependent Snowflake syntax including qualified wildcard expansion in aggregate and scalar functions. Decimal normalization reproduces Snowflake inference such as:

• 10.00 -> DECIMAL(2, 0)
• 10.10 -> DECIMAL(3, 1)
• 0.00100 -> DECIMAL(4, 3)
• 1.2300e2 -> DECIMAL(3, 0)

Validation

• DataFusion SQL unit tests: 92 passed
• aggregate and scalar planner integration tests
• qualified wildcard planner tests
• decimal literal planner integration tests
• cargo check -p datafusion
• Clippy for datafusion-common, datafusion-expr, datafusion-sql, and datafusion with -D warnings
• cargo fmt --all -- --check
• Taplo workspace formatting check
• full GitHub Actions matrix for the stacked PRs, including SQLite sqllogictests