sdf: Replace use of tbb::parallel_for with WorkParallelForTBBRange

Going through libWork adds support for clients with custom task
management systems. This drops the use of tbb::static_partitioner since
there currently is no matching concept defined in libWork. If there is
performance fallout, we could consider adding it to the API like we
did for the RangeType concept.

(Internal change: 2379721)

Author: sunyab

pxr/usd/sdf/crateData.cpp

@@ -28,8 +28,6 @@
 #include "pxr/usd/sdf/payload.h"
 #include "pxr/usd/sdf/schema.h"
 
-#include <tbb/parallel_for.h>
-
 #include <algorithm>
 #include <functional>
 #include <iostream>
@@ -815,7 +813,7 @@ class Sdf_CrateDataImpl
         }
 
         // Create all the specData entries and store pointers to them.
-        tbb::parallel_for(
+        WorkParallelForTBBRange(
             tbb::blocked_range<size_t>(0, specs.size()),
             [this, crateFile, &liveFieldSets, &specs](
                 tbb::blocked_range<size_t> const &r) {
@@ -831,8 +829,7 @@ class Sdf_CrateDataImpl
                         liveFieldSets.find(spec.fieldSetIndex)->second;
 
                 }
-            },
-            tbb::static_partitioner());
+            });
 
         _lastSet = _data.end();