Summary

motivation: #11581

need to Clarify where it can be used (may lead to diff impl)

snowflake supports mix usage of col_position and col_name when possible

select max($1), max(c) from variant_test where $1=true group by $1;

scopes to consider:

• statement: select stage only? or any statement?
• clauses： all (where, group by.. ) vs select only
• allow mixed usage or not

proposal:

1. all clauses and mixed usage when possible
   1. table and parquet
   2. json (only support $1) (. not sure if it is easy to impl mixed use)
2. CSV/TSV only support col_position