TPC-H benchmark can optionally write JSON output file with benchmark summary

[andygrove]

Which issue does this PR close?

Closes #1757.

Rationale for this change

To help with benchmark automation and reporting, I would like the benchmark results to be written to a JSON file.

What changes are included in this PR?

This PR adds a new --output argument to the tpch benchmark. When specified, a JSON summary file will be written to the specified directory, containing the benchmark results.

Example JSON output

{
  "benchmark_version": "5.0.0",
  "datafusion_version": "6.0.0",
  "num_cpus": 48,
  "start_time": 1644167292,
  "arguments": [
    "benchmark",
    "datafusion",
    "--iterations",
    "1",
    "--path",
    "/mnt/bigdata/tpch/sf100-tbl",
    "--format",
    "tbl",
    "--query",
    "1",
    "--batch-size",
    "4096",
    "-o",
    "/tmp"
  ],
  "query": 1,
  "iterations": [
    {
      "elapsed": 210781.71820099998,
      "row_count": 4
    }
  ]
}

Are there any user-facing changes?

There is a new --output option when running the tpch benchmarks.

[Dandandan]

@andygrove FYI I pushed a fix for a clippy linting error

[alamb]

Query 1 iteration 2 took 62666.3 ms and returned 4 rows
Query 1 avg time: 62579.50 ms
Writing summary file to /tmp/tpch-q1-1644431672.json

It is pretty neat:

alamb@MacBook-Pro-2 arrow-datafusion % cat /tmp/tpch-q1-1644431672.json 
{
  "benchmark_version": "5.0.0",
  "datafusion_version": "6.0.0",
  "num_cpus": 16,
  "start_time": 1644431672,
  "arguments": [
    "benchmark",
    "datafusion",
    "-o",
    "/tmp",
    "-p",
    "/Users/alamb/Software/tpch_data/SF1",
    "-q",
    "1",
    "--format",
    "tbl"
  ],
  "query": 1,
  "iterations": [
    {
      "elapsed": 62607.731700000004,
      "row_count": 4
    },
    {
      "elapsed": 62464.438148,
      "row_count": 4
    },
    {
      "elapsed": 62666.318697,
      "row_count": 4
    }
  ]

[alamb]

Follow on PR that I needed to test this more easily: #1800