Add seq-length argument to distribued BERT example

The --seq-length argument was missing from the example distributed BERT
pre-training script which throws an error while attempting to parse the
arguments as the code expects all arguments to be specified and have a
non-None value.
Signed-Off-By: Robert Clark <roclark@nvidia.com>

examples/pretrain_bert_distributed_with_mp.sh

@@ -23,6 +23,7 @@ python -m torch.distributed.launch $DISTRIBUTED_ARGS \
        --num-attention-heads 16 \
        --micro-batch-size 2 \
        --global-batch-size 16 \
+       --seq-length 512 \
        --max-position-embeddings 512 \
        --train-iters 1000000 \
        --save $CHECKPOINT_PATH \