Memory leak fixes in the connector

[taniabogatsch]

This PR builds on the work of #151. See also my comment here.

Changes in this PR:

• the Connector now exposes its Close method (done by @levakin). This PR contains the respective commits of #151.
  • explicitly calling c.(io.Closer).Close() is crucial to avoid memory leaks in the database object
  • exposing Close simplifies closing the Connector for users, reducing the risk of leaks and increasing usability
• maybe this is a lack in my understanding of Go, but I changed db *C.duckdb_database to db C.duckdb_database in the Connector because it is already a pointer (typedef struct _duckdb_database {void *__db;} * duckdb_database;). Or was there a reason to store it as a pointer?
• I added C.duckdb_destroy_config(&config) to more places to avoid leaks
• The following two places were leaking, as they never freed the C.CString, to the best of my knowledge
  • C.duckdb_set_config(config, C.CString("duckdb_api"), C.CString("go"))
  • C.duckdb_set_config(config, C.CString(k), C.CString(v[0]))
  • @levakin added setConfig, and I added explicit calls to C.free
• we do not have to keep the config-object alive after opening a DuckDB database

cc @marcboeker @levakin

[taniabogatsch]

This is an update to the main PR message concerning my most recent commit.

• A fix to prevent func (s *stmt) execute(ctx context.Context, args []driver.NamedValue) (*C.duckdb_result, error) from leaking memory on context errors (missing C.duckdb_destroy_result(&res)).
• A minor fix that prevents TestTypeNamesAndScanTypes from leaking memory.
• I added a section to the README.md that addresses memory allocations in go-duckdb. It explains why calling Close is crucial in go-duckdb to avoid memory leaks. While adding that section, I also expanded the readme file to make coding patterns more explicit.
• Some tidying

Note: there are still a few leaks in the appender, which @maiadegraaf addresses in #150.

[marcboeker]

Thanks @taniabogatsch