Range query scan_window steps by window_size_ms, not slide_interval_ms, for Sliding aggregations

[milindsrivastava1997]

execute_range_query_pipeline's per-step scan_window (mod.rs) steps the bucket_map by tumbling_window_ms/keys_tumbling_window_ms, both sourced from window_size_ms (finish_range_context, promql.rs). For WindowType::Sliding aggregations, buckets are actually persisted at slide_interval_ms granularity (precompute_engine/window_manager.rs:111, num_panes = window_size_ms / slide_interval_ms), not window_size_ms. When slide_interval_ms < window_size_ms, real buckets land on timestamps the scan never visits and are silently dropped.

Affects both the value side (pre-existing, predates #583/#587) and the keys side (introduced by #583, same root cause). Confirmed via 2 RED tests.

#[tokio::test(flavor = "multi_thread")]
async fn diagnostic_bug2_sliding_keys_bucket_off_window_size_grid() {
    // window_size_ms=2000, slide_interval_ms=1000. Delta bucket at
    // start=1000 (on the slide_interval_ms grid, NOT window_size_ms grid).
    let mut keys_add = DeltaSetAggregatorAccumulator::new();
    keys_add.add_key(KeyByLabelValues {
        labels: vec!["host-a".to_string(), "evt-1".to_string()],
    });

    let engine = create_range_engine_dual_input_sliding_keys(
        "event_frequency",
        AggregationType::CountMinSketch,
        AggregationType::DeltaSetAggregator,
        vec!["host", "event"],
        vec![(3000, None, Box::new(CountMinSketchAccumulator::new(2, 3)) as Box<dyn AggregateCore>)],
        vec![(1000, 2000, None, Box::new(keys_add) as Box<dyn AggregateCore>)],
        2000,
        1000,
        "count(event_frequency) by (host, event)",
    );

    let query = "count(event_frequency) by (host, event)";
    let result = engine.handle_range_query_promql(query.to_string(), 3.0, 3.5, 1.0);
    let (_, qr) = result.expect("range query failed");
    let elements = matrix_values(qr);

    assert!(
        key_has_sample_at(&elements, "host-a", 3000),
        "BUG CONFIRMED: host-a's delta bucket (start=1000, on the \
         slide_interval_ms=1000 grid but not the window_size_ms=2000 grid) \
         was NOT found -- scan_window is stepping the keys bucket_map by \
         window_size_ms instead of slide_interval_ms for this Sliding key \
         aggregation"
    );
}

#[tokio::test(flavor = "multi_thread")]
async fn diagnostic_bug2_sliding_values_bucket_off_window_size_grid() {
    // Same hypothesis, VALUE side. Sliding SumAccumulator aggregation,
    // window_size_ms=2000, slide_interval_ms=1000. Two buckets at start=0
    // (value=10.0) and start=1000 (value=5.0) -- both on the
    // slide_interval_ms grid, only one on the window_size_ms grid.
    let query = "sum_over_time(http_requests[1s])";
    let engine = create_engine_multi_timestamp_with_window(
        "http_requests",
        AggregationType::Sum,
        vec!["host"],
        vec![
            (1000, Some(vec!["host-a".to_string()]), Box::new(SumAccumulator::with_sum(10.0)) as Box<dyn AggregateCore>),
            (2000, Some(vec!["host-a".to_string()]), Box::new(SumAccumulator::with_sum(5.0)) as Box<dyn AggregateCore>),
        ],
        query,
        2000, // window_size_ms
        WindowType::Sliding,
    );
    // create_engine_multi_timestamp_with_window hardcodes slide_interval_ms:
    // 1000 internally, matching buckets at start=0 and start=1000.
    // step must be a multiple of window_size_ms (2000) or
    // validate_range_query_params rejects the query outright.
    let result = engine.handle_range_query_promql(query.to_string(), 2.0, 2.5, 2.0);
    let (_, qr) = result.expect("range query failed");
    let elements = matrix_values(qr);
    assert_eq!(elements.len(), 1, "expected one series for host-a");
    let value = elements[0].samples.first().map(|s| s.value);

    assert!(
        (value.unwrap_or(f64::NAN) - 15.0).abs() < 1e-9,
        "BUG CONFIRMED (pre-existing, value side): expected both buckets \
         (10.0 + 5.0 = 15.0) merged, got {:?} -- tumbling_window_ms is \
         stepping by window_size_ms=2000 instead of slide_interval_ms=1000",
        value
    );
}

Both FAIL against current code (keys test: bucket never found, elements empty; values test: merged value 10.0 instead of 15.0, second bucket silently dropped).