docs: add window function frame specification examples to SQL and DataFrame pages

SQL page: added ROWS BETWEEN examples (rolling average, running total)
with the full syntax (UNBOUNDED PRECEDING, N PRECEDING, CURRENT ROW,
N FOLLOWING, UNBOUNDED FOLLOWING).

DataFrame page: added rolling aggregate example with frame property
and documented frame bound values.

Author: teamchong

docs/src/content/docs/dataframe-api.mdx

@@ -99,6 +99,21 @@ df.window({
 
 Supported: `row_number`, `rank`, `dense_rank`, `lag`, `lead`, rolling `sum`/`avg`/`min`/`max`/`count` with frame specification.
 
+Rolling aggregates with frame:
+
+```typescript
+df.window({
+  fn: "avg",
+  column: "amount",
+  partitionBy: ["region"],
+  orderBy: [{ column: "created_at", direction: "asc" }],
+  alias: "rolling_avg",
+  frame: { type: "rows", start: -2, end: 0 }, // 3-row trailing window
+})
+```
+
+Frame bounds: numeric offsets (negative = preceding, positive = following), `"unbounded"`, or `"current"`.
+
 ## Computed columns
 
 ```typescript

docs/src/content/docs/sql.mdx

@@ -101,6 +101,20 @@ SELECT name, LAG(amount, 1) OVER (ORDER BY created_at) AS prev_amount FROM order
 
 `ROW_NUMBER`, `RANK`, `DENSE_RANK`, `LAG`, `LEAD`, and rolling `SUM`/`AVG`/`MIN`/`MAX`/`COUNT` with frame specification.
 
+Frame specification supports `ROWS` and `RANGE` with `UNBOUNDED PRECEDING`, `N PRECEDING`, `CURRENT ROW`, `N FOLLOWING`, `UNBOUNDED FOLLOWING`:
+
+```sql
+-- Rolling 3-row average
+SELECT name, AVG(amount) OVER (
+  ORDER BY created_at ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING
+) AS rolling_avg FROM orders
+
+-- Running total
+SELECT name, SUM(amount) OVER (
+  PARTITION BY region ORDER BY created_at ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW
+) AS running_total FROM orders
+```
+
 ### Common Table Expressions (CTEs)
 
 ```sql