Expose built_in columns using ds.field() in datasets

### Describe the enhancement requested

For all dataset/scanner functions like to_table(), head(), etc.. which support @columns..

This works..
`to_table(columns=["my_column", "__filename"])`

But built-in columns are not supported using expressions:
`to_table(columns = {"my_column": ds.field("my_column") , "filename": ds.field("__filename")})`

I'm trying to create an expressions to parse out YYYYMMDD from the following filename:
VAR_pnl.20230829.csv.gz
But I need to be able to reference the built-in column "__filename" in the expression..

My only other option is to extract everything into memory first using @columns= ["a","b", "__filename"], but this isn't ideal for memory management.

### Component(s)

Python