fix(sqlite): columns with the same name stay separate in query results (#1144)

[ZhuchkaTriplesix]

Closes #1144

Problem

SQLite query results were read as maps keyed by column name, so SELECT o.id, u.id showed one column with the wrong value. The same happened to SELECT 1, 1, o.*, u.* and the table browser's custom SQL.

Change

• sqliteUniqueColumnsSql (in sqlite_sql.dart) wraps a single SELECT or WITH statement in SELECT * FROM (...). SQLite then names the repeats id and id:1.
• The SQL editor and MCP path (SqliteSqlExecutionDelegate.executeQuery) apply it before the LIMIT is injected, so the LIMIT stays on the outer query.
• The table browser's custom SQL (loadCustomSql) applies it too.
• Other statements, several statements at once, and results with unique names are unchanged.

Behaviour to know about

The repeated columns are shown as id and id:1, which is how SQLite names them inside a subquery. Earlier, the repeat was lost entirely. sqflite does not expose the column list positionally, so this is the approach that keeps every column without a driver change.

Tests

test/core/database/sqlite_unique_columns_test.dart:

• unit cases for the wrapper (WITH, trailing comments and semicolons, statements left unchanged);
• regression on a real SQLite file through McpQueryService: o.id, u.id, SELECT 1, 1, o.*, u.*, a WITH query, and unique names keeping their order.

[github-actions]

Grid scroll benchmark

metric	base	PR	change
p50	16.86 ms	17.01 ms	+0.9%
p90	23.41 ms	25.03 ms	+6.9% ⚠️
p99	51.45 ms	53.14 ms	+3.3%
stutters	290.00	267.00	-7.9%

Informational only (threshold 5%). Shared CI runners are noisy; re-run before trusting a single result.