'orc.column.index.access'='false' for athena table

No, this shouldn't affect performance if configured either way, basically one just does an index lookup and the other uses a map lookup by name and this is performed at a high level once per file (relevant Presto source - https://github.com/prestodb/presto/blob/6647e13f64883f7cfa89221d91b981bcc3a57618/presto-hive/src/main/java/com/facebook/presto/hive/HiveUtil.java#L976)

More interesting is the functional difference. I'm curious why you set this to false to add a column? You can add a column with it set to true. Generally, having this set to true (map by name) is easier to manage. The one area that having it set to false would be useful is if you wanted to rename a column without having to rewrite the data since it's not bound to the name but rather the index. Problem areas though that you have to write data in the same order as the metadata definition even if a column is removed. Generally for most schema evolution use cases (added or removing columns) mapping by name is easier to work with. Some discussion about this in the Presto repo since actually in standard Presto the default is false (index lookup) but Athena internally has its own flags for this it's managing (with different defaults) - https://github.com/prestodb/presto/issues/8911