Does anyone know whether it’s possible to emulate a graph database with a relational by adding specific meta-data tables or indices? Emulate in the sense that you can do efficient graph traversals that drill along a path for a certain non-trivial depth.
How do sql query engines backed by S3 like Trino (Athena) behave for euch queries?
I don’t know. But recursive CTEs habe been part of the standard since SQL:1999, so if they advertise compliance with SQL:1999 or a newer version of the standard, they should at least be able to execute the query. To find out whether they do so efficiently, you could do an experiment (remember most SQL engines allow you to look at the query plan for a given query) or read their documentation.