export from duckdb

[tarasglek]

https://duckdb.org/docs/data/json/sql_to_and_from_json.html supports wonderful constructs like

COPY (SELECT 42 AS a, 'hello' AS b) TO 'query.json'

and others like https://duckdb.org/docs/sql/statements/copy.html#copy--

After outputting to such files can do:
SELECT * FROM glob('*'); to list em;

copyFileToBuffer to load them into JS and offer for download to users.

We can pattern match whether there seems to be a copy statement and run glob(*) before/after to get a diff of files to offer to download

Algorithm:

1. pattern match to see if query is `/^\s*\bCOPY\b.*\bTO\b/i
2. run glob(*) to get initial list of files
3. run query
4. run glob(*) to find new files

Followups:

• We should also support dropping them to reduce memory.
• We should also support loading attachments into them.