sql: support unicode escapes

The following syntax is supported in postgres, but in crdb results in a syntax error:

```
SELECT U&'d\0061t\+000061' AS U&"d\0061t\+000061";
-- postgres: data

SELECT U&'d!0061t\+000061' UESCAPE '!' AS U&"d*0061t\+000061" UESCAPE '*';
-- postgres: dat\+000061

SELECT U&'a\\b' AS "a\b";
-- postgres: a\b

SELECT E'\\x De Ad Be Ef '::bytea;
-- postgres: \xdeadbeef
```

Found in pg_regress strings

Jira issue: CRDB-33912

Start by locating the pg_regress strings referenced in the issue and run the listed SQL examples against CockroachDB to reproduce the syntax errors. Compare the results with the PostgreSQL comments; done means the Unicode and escape syntaxes parse and produce the expected results, including the bytea value.

Written by the indexing model from the issue text.