support function: current_timestamp, current_date, now

same functions in Spark SQL.

1. There is `unix_timestamp("")` as a temp solution, not convenient though. unix_timestamp returns unix epouch time in seconds as int64, one must write `timestamp (unix_timestamp("") * 1000)` to archive current_timestamp.
2. all `current_*` function should executed at begining of one SQL, means all `current_*` functions calls in same SQL should guarantee return the same value, no matter where the `current_*` call is.

Look for existing timestamp functions like `unix_timestamp` in the SQL function implementation. The new functions `current_timestamp`, `current_date`, and `now` need to be added, ensuring they return the same value for all calls within a single SQL statement. Check how query execution context is managed to guarantee consistency. Test by writing SQL queries that call these functions multiple times.

Rédigé par le modèle d'indexation à partir du texte de l'issue.