Add optional decimal literal normalization

[osipovartem]

Which issue does this PR close?

Part of Embucket Snowflake query compatibility work.

Rationale for this change

Snowflake removes insignificant trailing zeros when inferring exact numeric literal types: 10.00 is NUMBER(2,0), 10.10 is NUMBER(3,1), and 0.00100 is NUMBER(4,3). DataFusion currently preserves the textual scale. This affects result metadata and decimal rendering in otherwise-correct queries.

What changes are included in this PR?

• Add opt-in datafusion.sql_parser.trim_decimal_literal_trailing_zeros (default false).
• Normalize only positive decimal scale; scientific notation that requires negative scale remains intact.
• Preserve a leading zero in precision for values below one, matching Snowflake exact numeric inference.
• Add parser and planner coverage and document the option.

The default DataFusion behavior is unchanged. Rustice enables this option explicitly.

Are there any user-facing changes?

Only for applications that enable the new option.

How was this tested?

• cargo test -p datafusion-sql --lib (92 passed)
• cargo test -p datafusion-sql --test sql_integration parse_decimals_trim_insignificant_trailing_zeros
• cargo check -p datafusion
• cargo clippy -p datafusion-common -p datafusion-sql -p datafusion --all-targets -- -D warnings
• cargo fmt --all -- --check