gh-158277: sqlite3 CLI completer: include views in column completion

[Tony123-tech]

Fixes gh-158277.

[Tony123-tech]

Thanks for the review! I've fixed both issues:

1. Added _col_view to the _table definition
2. Changed the decorator to the correct SQLite version (3.16.0 for pragma_table_xinfo)

[tanloong]

1. The extra _col_view_alias column check in test_complete_view_columns() provides no more infomation than _col_table_alias. Asserting _col_table_alias alone in the candidates is enough for test purpose. How about removing the _col_view_alias and _col_view definition:

CREATE TABLE _table (_col_table);
CREATE VIEW _view AS SELECT _col_table AS _col_table_alias FROM _table;

2. The @unittest.skipIf(sqlite3.sqlite_version_info < (3, 16, 0), ...) decorator, which is originally for test_complete_functions(), is occupied by the newly added test_complete_view_columns() in this PR. How about replace the test_complete_view_columns() with new tests in the already existing test_complete_columns() like this:

    def test_complete_columns(self):
        input_ = textwrap.dedent("""\
            CREATE TABLE _table (_col_table);
            CREATE TEMP TABLE _temp_table (_col_temp);
+           CREATE VIEW _view AS SELECT _col_table AS _col_table_alias FROM _table;
            ATTACH ':memory:' AS attached;
            CREATE TABLE attached._attached_table (_col_attached);

            SELECT _col_\t\tta\tFROM _table;
            .quit\n""").encode()
        output = self.write_input(input_)
        lines = output.decode().splitlines()
        indices = [
            i for i, line in enumerate(lines) if line.startswith(self.PS1)
        ]
        start, end = indices[-3], indices[-2]
        candidates = [l.strip() for l in lines[start+1:end]]

        self.assertEqual(
-           candidates, ["_col_attached", "_col_table", "_col_temp"]
+           candidates, ["_col_attached", "_col_table", "_col_table_alias", "_col_temp"]
        )

[tanloong]

Looks good to me. Let's wait for erlend-aasland's review as he is the module maintainer, he might ask for some changes, in which case I'm sorry if I suggested something wrong.

[Tony123-tech]

@tanloong @erlend-aasland The "Unresolved review" check is failing
because the two review comments from 2 weeks ago are still open. I've
addressed both points in 0d647bb (the comments are marked outdated).
Could one of you resolve them, or let me know if there's something
else that needs to change?

[encukou]

Looks good to me; I'll merge ~Friday if there are no objections.