Preserve Snowflake PIVOT column names

[osipovartem]

Summary

• preserve the SQL literal text, including string quotes, in generated PIVOT column names
• honor explicit PIVOT value aliases and normal identifier normalization
• use the resolved name consistently in the aggregate and DEFAULT ON NULL projection

Snowflake names a column generated from IN ('2023_Q1') as '2023_Q1', so it is addressed as "'2023_Q1'". The previous planner stripped the single quotes and produced 2023_Q1.

Validation

• live Snowflake oracle query confirmed the '2023_Q1' output name and explicit alias behavior
• cargo +1.94.0 test -p datafusion-sql --offline - 92 unit tests, 582 integration tests, and 12 doctests passed (1 doctest ignored)
• cargo +1.94.0 clippy -p datafusion-sql --all-targets --offline -- -D warnings
• cargo +1.94.0 fmt --all -- --check
• git diff --check

Dependency

This PR is based on #47, which consolidates the existing DF55 compatibility commits into embucket-sync-df55.0.0. After #47 merges, this PR can be retargeted to that branch without carrying unrelated changes.