[Spark] Expose a write summary from VortexOutputWriter #11125

Problem or motivation

VortexOutputWriter.close() finishes the file but throws away the VortexWriteSummary that VortexWriter.finish() returns. To learn anything about the file it just wrote, a caller has to open it again.

Proposed solution

After a successful close(), expose a summary of the write:

• file size and row count, from VortexWriter.finish();
• number of batches;
• peak Arrow bytes of one batch;
• peak bufferedBytes() reported by the native writer.

Additional context

• Callers can collect file stats from tasks (sizes, rows per file) to check layout and skew, without listing and reopening files.
• Peak batch and buffered bytes show writer memory per file. A regression in native memory retention, like JNI writer keeps imported Arrow batches alive until the file finishes #11068, then shows up as a number that changes, not as OOM kills.