\\udc81 surrogates not allowed

https://meltano.slack.com/archives/CMN8HELB0/p1675083559152229

See if this character works with target-postgres. I believe we have a test that tests a lot of combinations of oddball UTF-8 characters. I believe \udc81 is a utf-8 char but I'm not sure will have to look.

Took a look and if you add this change

visch@DESKTOP-9BDPA9T:~/git/target-postgres$ git diff
diff --git a/target_postgres/tests/data_files/encoded_strings.singer b/target_postgres/tests/data_files/encoded_strings.singer
index 6e4c9c9..2407d77 100644
--- a/target_postgres/tests/data_files/encoded_strings.singer
+++ b/target_postgres/tests/data_files/encoded_strings.singer
@@ -29,4 +29,5 @@
 {"type": "RECORD", "stream": "test_strings_in_arrays", "record": {"id": 4, "strings": ["\u006D", "\u0101", "\u0199"]}}
 {"type": "RECORD", "stream": "test_strings_in_arrays", "record": {"id": 5, "strings": ["aaa", "Double quoting: \\u0041 \\u0001"]}}
 {"type": "RECORD", "stream": "test_strings_in_arrays", "record": {"id": 6, "strings": ["bbb", "Control Characters in string: \u0041 \u0001"]}}
+{"type": "RECORD", "stream": "test_strings_in_arrays", "record": {"id": 7, "strings": ["bbb", "Control Characters in string: \u0041 \u0001 \udc81"]}}
 {"type": "STATE", "value": {"test_strings": 11, "test_strings_in_objects": 11, "test_strings_in_arrays": 6}}

And run the related test psycopg will spit out:

E           sqlalchemy.exc.DataError: (psycopg2.errors.InvalidTextRepresentation) invalid input syntax for type json
E           LINE 1: ...n string: A \u0001"']::JSONB[]),(7, ARRAY['"bbb"','"Control ...
E                                                                        ^
E           DETAIL:  Unicode low surrogate must follow a high surrogate.
E           CONTEXT:  JSON data, line 1: ...
E
E           [SQL: INSERT INTO temp_test_strings_in_arrays_94e1245a_f2e0_4a0d_b6c7_33ff224f37e6 (id, strings) VALUES (%(id)s, %(strings)s::JSONB[])]
E           [parameters: ({'id': 1, 'strings': ['"simple string"', '"\\u03b1\\u03c0\\u03bb\\u03ae \\u03c3\\u03c5\\u03bc\\u03b2\\u03bf\\u03bb\\u03bf\\u03c3\\u03b5\\u03b9\\u03c1\\u03ac"', '"\\u7b80\\u5355\\u7684\\u5b57\\u4e32"']}, {'id': 2, 'strings': ['"cha\\u00eene simple"', '"quoted \\"string\\""']}, {'id': 3, 'strings': ['"various \\" \\\\ / \\n escape sequences"']}, {'id': 4, 'strings': ['"m"', '"\\u0101"', '"\\u0199"']}, {'id': 5, 'strings': ['"aaa"', '"Double quoting: \\\\u0041 \\\\u0001"']}, {'id': 6, 'strings': ['"bbb"', '"Control Characters in string: A \\u0001"']}, {'id': 7, 'strings': ['"bbb"', '"Control Characters in string: A \\u0001 \\udc81"']})]
E           (Background on this error at: https://sqlalche.me/e/14/9h9h)

.venv/lib/python3.8/site-packages/psycopg2/extras.py:1299: DataError

I think the way to allow this is to allow us to override this type to save this data in a binary format in Postgres instead of text. We'd need to allow overrides to data types (Meltano does allow this via a schema override) and we'd need to add support for BLOBs, and a mapping to JSON Schema somewhere

Start with target_postgres/tests/data_files/encoded_strings.singer and the related test that loads its oddball UTF-8 cases; reproduce the failure by adding the \udc81 record. Then review how schema overrides are handled and where PostgreSQL type mappings are defined. The scope is complete only after the project agrees on the override, binary/BLOB support, and JSON Schema mapping needed for this data.