Trino query runner: get_schema() ignores schema configuration, fetches all schemas

[MinhyukK0]

Issue Summary

When configuring a Trino data source with both catalog and schema specified, the schema browser still shows tables from all schemas in the catalog, not just the configured one.

This is because get_schema() only uses the catalog configuration but ignores schema:

# current code in redash/query_runner/trino.py
query = f"""
    SELECT table_schema, table_name, column_name, data_type
    FROM {catalog}.information_schema.columns
    WHERE table_schema NOT IN ('pg_catalog', 'information_schema')
"""

Problem

For large catalogs (e.g., Iceberg with Glue metastore), querying information_schema.columns without a schema filter causes:

1. Timeout/partial results — Glue API latency makes the full query exceed the trino Python client's default request_timeout (30s), returning incomplete results
2. Unusable schema browser — Even when results return, showing all schemas makes the browser noisy when users only need one schema

Related: #6059

Proposed Fix

When schema is configured, add it as a WHERE condition:

def get_schema(self, get_stats=False):
    if self.configuration.get("catalog"):
        catalogs = [self.configuration.get("catalog")]
    else:
        catalogs = self._get_catalogs()

    schema_filter = self.configuration.get("schema")

    schema = {}
    for catalog in catalogs:
        query = f"""
            SELECT table_schema, table_name, column_name, data_type
            FROM {catalog}.information_schema.columns
            WHERE table_schema NOT IN ('pg_catalog', 'information_schema')
        """
        if schema_filter:
            query += f" AND table_schema = '{schema_filter}'"

        results, error = self.run_query(query, None)
        ...

Environment

• Redash 25.8.0 (also confirmed in latest v26.3.0 source)
• Trino with Iceberg connector + AWS Glue catalog
• 15 schemas, ~105 tables, ~4894 columns

[MinhyukK0]

Opened a fix PR: #7681

The change adds a table_schema filter to the information_schema.columns query in get_schema() when the schema configuration is set. This narrows the query scope and prevents timeouts on large catalogs.