Add count and count_distinct select aggregates

[aisrael]

Summary

This branch extends REPL select() aggregates with count(:column) (non-null values) and count_distinct(:column) (distinct non-null values), alongside the existing sum, avg, min, and max forms. They work for global summaries (select-only aggregates) and for grouped pipelines after group_by().

Implementation

• Pipeline spec and DataFusion: New SelectItem::Count and SelectItem::CountDistinct variants wired through apply_select_spec_to_dataframe using DataFusion aggregate helpers.
• REPL planning: Parsing, validation, and display updated so these names are treated like other single-column aggregates.
• Docs: README and docs/REPL.md describe the new functions and distinguish read(...) |> count() (row count) from count(:col) inside select().
• Tests: Gherkin scenarios in features/repl/aggregates.feature and Rust unit tests cover global and grouped usage.

Made with Cursor