support remove_input_padding for BertForSequenceClassification models

[Altair-Alpha]

Related issue: #1755
Content:

1. Support remove_input_padding for BertForSequenceClassification models (implementation details given in code comment)
2. Refinement of the build.py script, e.g., the original script doen't have a input model dir parameter, and will init the model with random weights, which is not intuitive.
3. Since the model input is changed from 3 to 5, i.e. input_ids, input_lengths, token_type_ids, position_ids, max_input_length, I add a standalone run_remove_input_padding.py demo script, and show how to build them with only input_ids and token_type_ids.

I only implemented and tested this for BertForSequenceClassification but not other BERT models yet, please feel free to do further work on this :)

[Altair-Alpha]

For our data, there's about 35% latency and 28% throughput enhancement when batch_size is 16, and the diff of output logits between the pytorch model and trt with remove_input_padding is below 0.01

[nv-guomingz]

Thanks @Altair-Alpha, we'll merge your changes and upstream to github next week.

[nv-guomingz]

hi @Altair-Alpha , we've merged your PR into our internal code base with several minimal changes for internal ci.
This MR will be avaiable on next Tensor Tuesday.