Merge pull request #639 from RedisAI/fix_leak_in_running_onnx_with_batching

Fix leak in running onnx with batching

Author: alonre24

src/backends/onnxruntime.c

@@ -571,6 +571,7 @@ int RAI_ModelRunORT(RAI_ModelRunCtx **mctxs, RAI_Error *error) {
                 ONNX_VALIDATE_STATUS(ort->GetDimensionsCount(info, &ndims))
                 int64_t dims[ndims];
                 ONNX_VALIDATE_STATUS(ort->GetDimensions(info, dims, ndims))
+                ort->ReleaseTensorTypeAndShapeInfo(info);
                 if (dims[0] != total_batch_size) {
                     RAI_SetError(error, RAI_EMODELRUN,
                                  "ERR Model did not generate the expected batch size");