Fix allocation failures during sorting with spill-to-disk

This change replaces `try_resize` with `resize` in three sites, allowing memory
to overshoot the configured pool size. These are sites where we don't fall back
to spilling to disk when the allocation fails.

Fixes: apache#12136

Author: cjolowicz

datafusion/physical-plan/src/sorts/builder.rs

@@ -69,7 +69,8 @@ impl BatchBuilder {
 
     /// Append a new batch in `stream_idx`
     pub fn push_batch(&mut self, stream_idx: usize, batch: RecordBatch) -> Result<()> {
-        self.reservation.try_grow(batch.get_array_memory_size())?;
+        // Allow memory to exceed the limit
+        self.reservation.grow(batch.get_array_memory_size());
         let batch_idx = self.batches.len();
         self.batches.push((stream_idx, batch));
         self.cursors[stream_idx] = BatchCursor {

datafusion/physical-plan/src/sorts/sort.rs

@@ -580,7 +580,8 @@ impl ExternalSorter {
         if self.runtime.disk_manager.tmp_files_enabled() {
             let size = self.sort_spill_reservation_bytes;
             if self.merge_reservation.size() != size {
-                self.merge_reservation.try_resize(size)?;
+                // Allow memory to exceed the limit
+                self.merge_reservation.resize(size);
             }
         }