As I’m working with my data, setting the formats, I’m finding most of the character I could use as delimited are actually used in the remarks - even the tilde, in some of them.
Other ETL tools can actually let you set the Quote equivalent to None since the content of a Tab delimited field, or indeed a CSV, may not have encapsulated string.
I’m concerned that picking a surrogate character e.g using a caret or a curly bracket may end up being entered by users in the data.