Bug with tuple

I'm running this:

```go
func (i *snapshotIterator) Next() ([]map[string]any, error) {
if i.done {
return nil, fmt.Errorf("no more rows to scan")
}

if i.iter == nil {
colNames := make([]string, len(i.columns))
for idx, col := range i.columns {
// Cassandra identifiers are typically lowercase; avoid quoting to keep it simple.
colNames[idx] = col.Name
}

// Build fully-qualified table name keyspace.table
stmt := fmt.Sprintf("SELECT %s FROM %s.%s", strings.Join(colNames, ","), i.keyspace, i.tableName)
fmt.Println("stmt", stmt)
i.iter = i.session.Query(stmt).PageSize(i.batchSize).Iter()
}

var rows []map[string]any
for len(rows) < i.batchSize {
row := make(map[string]any)
if !i.iter.MapScan(row) {
i.done = true
break
}

out, err := i.normalizeRow(row)
if err != nil {
return nil, fmt.Errorf("failed to normalize row: %w", err)
}

rows = append(rows, out)
}

return rows, nil
}
```

You can clearly see that `c_tuple_int_text` is null:

However, this is what I'm getting back:

```go
stmt SELECT id,c_ascii,c_bigint,c_blob,c_boolean,c_date,c_decimal,c_double,c_float,c_frozen_list_int,c_inet,c_int,c_smallint,c_text,c_time,c_timestamp,c_timeuuid,c_tinyint,c_tuple_int_text,c_uuid,c_varchar,c_varint,c_list_int,c_map_text_int,c_set_int FROM artie_test.data_type_test
key c_double value 4.14159
key c_time value 12h34m56s
key c_timestamp value 2025-09-10 23:45:22.811 +0000 UTC
key c_tinyint value 42
key c_varint value 1234567890123456789
key c_timeuuid value 37a76f2c-8ea0-11f0-bed0-d6c22f62521e
key c_list_int value [10]
key c_map_text_int value map[a:1 b:2]
key c_set_int value [1]
key c_ascii value ascii_1
key c_boolean value false
key c_decimal value 123.45
key c_float value 3.71828
key c_tuple_int_text[0] value 0
key c_tuple_int_text[1] value
key c_varchar value varchar_1
key id value 2
key c_blob value [1 2 3 1]
key c_frozen_list_int value [1]
key c_inet value 10.0.1.2
key c_int value 1001
key c_smallint value 1234
key c_text value text_1
key c_uuid value baed9c37-e600-497d-857d-852d395dad07
key c_bigint value 9000000001
key c_date value 2025-09-10 00:00:00 +0000 UTC
```

Start at snapshotIterator.Next and inspect normalizeRow, since the report shows the tuple value changing during row normalization but names no source file or test. Trace how MapScan represents a null c_tuple_int_text value and verify that normalization preserves null rather than producing zero values; done means the returned row matches the Cassandra value.

Written by the indexing model from the issue text.