Support Snowflake MERGE ALL BY NAME

[osipovartem]

Summary

• add dedicated AST variants for Snowflake MERGE ... UPDATE ALL BY NAME and INSERT ALL BY NAME
• preserve explicit UPDATE SET assignments as a typed MergeUpdateKind::Set
• retain spans, formatting, serde/visitor support, and existing BigQuery/common parser coverage
• add Snowflake parse and display round-trip coverage

Why

Snowflake maps ALL BY NAME by column name rather than ordinal position. Representing it as a dedicated AST kind prevents downstream planners from confusing it with wildcard or empty-assignment semantics.

Validation

• cargo +1.95.0 test --all-targets
• cargo +1.95.0 check --all-targets
• cargo +1.95.0 fmt --all -- --check