Creating glue catalog tables using crawler and then reading tables from glue job versus directly reading from S3 using glue job

Question

To read from S3, transform the data, and then write the output to S3 we can create glue catalog tables using glue crawler and then read the tables from the Glue job or we can directly read from S3 using the glue job.

What advantages and disadvantages does these two approaches have over the other?

One advantage of the first approach is that the tables created using crawler can be accessed from Athena and we can run athena queries for analysis.

Thank you.