Basically good, I think. I have modified your test a bit, testing more directly for the presence of the LATIN-1 encoded character and the absence of the UTF-8 encoded character, by reading in the file with pg_read_binary_file, and adding a test for implicit encoding by setting client_encoding.
The revised tests look better to me. Checking the raw bytes with pg_read_binary_file() directly verifies that LATIN1 output does not contain
the UTF-8 sequence, and the added implicit client_encoding case too
