llama2 -70b support and calibration

[arekay-nv]

• Ensure that llama2-70b executes end-to-end and reports number statistically similar to legacy loadgen as submitted in MLPerf inference v5.1 for all scenarios.
• Add documentation to enable reproducing the results.

[arekay-nv]

@tianmu-li - can this be closed?

[tianmu-li]

@tianmu-li - can this be closed?

Yes, handled by #59 and #97

[arekay-nv]

Thanks @tianmu-li