benchmark_main should report I/O flop count

Currently benchmark_main only reports internal flop count with output akin to the following for a 5 stage pipeline.

```
Pipeline:
nodes: 266, delay: 72ps
[Stage 0] flops: 706 ( 0 dups, 0 constant)
nodes: 140, delay: 69ps
[Stage 1] flops: 556 ( 0 dups, 0 constant)
nodes: 16, delay: 19ps
[Stage 2] flops: 556 ( 0 dups, 0 constant)
nodes: 0, delay: 0ps
[Stage 3] flops: 556 ( 0 dups, 0 constant)
nodes: 10, delay: 40ps
Total pipeline flops: 2374 (0 dups, 0 constant)
```

This issue is to enhance it so it (optionally) reports input or output flopping

```
Pipeline:
[Stage 0] flops: 100 ( 0 dups, 0 constant)
nodes: 266, delay: 72ps
[Stage 1] flops: 706 ( 0 dups, 0 constant)
nodes: 140, delay: 69ps
[Stage 2] flops: 556 ( 0 dups, 0 constant)
nodes: 16, delay: 19ps
[Stage 3] flops: 556 ( 0 dups, 0 constant)
nodes: 0, delay: 0ps
[Stage 4] flops: 556 ( 0 dups, 0 constant)
nodes: 10, delay: 40ps
[Stage 5] flops: 100 ( 0 dups, 0 constant)
Total pipeline flops: 2574 (0 dups, 0 constant)
```