A. Enable Snowpipe's 'ERROR_INTEGRATION' , examine the error logs for malformed JSON records, and adjust the COPY INTO statement with appropriate FILE_FORMAT options to handle the nested arrays.

B. Implement a pre-processing step using a Snowflake task to validate the JSON data before loading it into the 'ORDERS' table using a COPY INTO statement. The Task would filter out bad records.

C. Use a 'VALIDATE' statement with the same COPY INTO statement to identify records that will fail. Then, modify the COPY INTO statement to handle the errors or exclude the problematic records using a 'WHERE clause in the transformation logic of the COPY INTO statement or inline SQL Transformations

D. Increase the compute resources allocated to the virtual warehouse used by the Snowpipe. Also, disable Snowpipe and load the data manually using the COPY INTO command to identify any errors during load.

E. Enabling the ERROR=CONTINUE parameter on the COPY INTO statement used in the pipe, and regularly querying the function to identify issues in the loaded data.