pg_stat_statements: Add rows_scanned and rows_filtered columns

[NikolayS]

pg_stat_statements: Add rows_scanned and rows_filtered columns

Summary

This PR adds two new columns to pg_stat_statements to help identify queries that may benefit from better indexing:

• rows_scanned - Total rows fetched from storage by scan nodes before any filtering
• rows_filtered - Total rows removed by filter conditions (scanqual, joinqual, and other quals)

These metrics are useful for identifying inefficient queries that scan many rows but return few results.

Implementation Details

• Enables per-node instrumentation (INSTRUMENT_ALL) in ExecutorStart hook
• Walks the plan tree in ExecutorEnd to collect statistics from all nodes
• For rows_scanned: sums tuples from scan nodes (SeqScan, IndexScan, etc.) including filtered tuples
• For rows_filtered: sums nfiltered1 (scanqual/joinqual) and nfiltered2 (other quals) from all nodes
• Reads both ntuples and tuplecount from instrumentation to capture complete tuple counts

Example Output

SELECT query, rows, rows_scanned, rows_filtered 
FROM pg_stat_statements;

                 query                  | rows | rows_scanned | rows_filtered
----------------------------------------+------+--------------+---------------
 SELECT * FROM test_rows WHERE id <= $1 |   10 |         1000 |           990
 SELECT * FROM test_rows WHERE id > $1  |  500 |         1000 |           500
 SELECT * FROM test_simple              |  100 |          100 |             0
 SELECT * FROM test_simple LIMIT $1     |    5 |            5 |             0