add comment about the accumulator usage

Signed-off-by: Zach Puller <zpuller@nvidia.com>

sql-plugin/src/main/scala/org/apache/spark/sql/rapids/GpuTaskMetrics.scala

@@ -214,6 +214,11 @@ class GpuTaskMetrics extends Serializable {
   def updateMaxGpuMemory(taskAttemptId: Long): Unit = {
     val maxMem = RmmSpark.getAndResetGpuMaxMemoryAllocated(taskAttemptId)
     if (maxMem > 0) {
+      // This metric tracks the max amount of memory that is allocated on the gpu during
+      // the lifespan of a task. However, this update function only gets called once on task
+      // completion, whereas the actual logic tracking of the max value  during memory allocations
+      // lives in the JNI. Therefore we can stick the convention here of calling the add method
+      // instead of adding a dedicated max method to the accumulator.
       maxDeviceMemoryBytes.add(maxMem)
     }
   }