perf: add benchmarks for importing all packages

[anonrig]

Related to nodejs/undici#1679

➜  fast-querystring git:(perf/benchmark) npm run benchmark:import

> fast-querystring@1.0.0 benchmark:import
> node benchmark/import.mjs

╔═════════════════════════════╤═════════╤══════════════════╤═══════════╗
║ Slower tests                │ Samples │           Result │ Tolerance ║
╟─────────────────────────────┼─────────┼──────────────────┼───────────╢
║ @aws-sdk/querystring-parser │    1000 │  10219.42 op/sec │  ± 0.47 % ║
║ querystringparser           │    1000 │  10733.12 op/sec │  ± 0.47 % ║
║ query-string                │    1000 │  11056.24 op/sec │  ± 0.47 % ║
║ qs                          │    1000 │  11117.31 op/sec │  ± 0.49 % ║
║ querystringify              │    1000 │  11737.47 op/sec │  ± 0.49 % ║
║ fast-querystring            │    1500 │  37799.39 op/sec │  ± 0.86 % ║
╟─────────────────────────────┼─────────┼──────────────────┼───────────╢
║ Fastest test                │ Samples │           Result │ Tolerance ║
╟─────────────────────────────┼─────────┼──────────────────┼───────────╢
║ node:querystring            │   10000 │ 228785.04 op/sec │  ± 1.19 % ║
╚═════════════════════════════╧═════════╧══════════════════╧═══════════╝

[climba03003]

Pretty unfair to compare from relative path and module name.
It give a false result to indicate this module is significant faster than others.

> fast-querystring@1.0.0 benchmark:import
> node benchmark/import.mjs

╔═════════════════════════════╤═════════╤══════════════════╤═══════════╗
║ Slower tests                │ Samples │           Result │ Tolerance ║
╟─────────────────────────────┼─────────┼──────────────────┼───────────╢
║ @aws-sdk/querystring-parser │    1000 │   1847.15 op/sec │  ± 0.46 % ║
║ querystringparser           │    1000 │   2005.94 op/sec │  ± 0.57 % ║
║ qs                          │    1000 │   2089.31 op/sec │  ± 0.29 % ║
║ fast-querystring            │    1000 │   2104.44 op/sec │  ± 0.28 % ║
║ querystringify              │    1000 │   2302.10 op/sec │  ± 0.58 % ║
║ query-string                │    1000 │   2302.31 op/sec │  ± 0.27 % ║
╟─────────────────────────────┼─────────┼──────────────────┼───────────╢
║ Fastest test                │ Samples │           Result │ Tolerance ║
╟─────────────────────────────┼─────────┼──────────────────┼───────────╢
║ node:querystring            │   10000 │ 118867.90 op/sec │  ± 1.15 % ║
╚═════════════════════════════╧═════════╧══════════════════╧═══════════╝

[anonrig]

@climba03003 This is already a pretty unfair comparison since node:querystring loads faster due to node internals. Can you open a pull request to update your changes?

[climba03003]

@climba03003 This is already a pretty unfair comparison since node:querystring loads faster due to node internals. Can you open a pull request to update your changes?

Yes, for comparing internal is unfair to others. At least, it could be fair the packages in node_modules.
I cannot find a good way to install the package itself without listing in dependency.
Would it be fine in listing itself in the devDependency?

[anonrig]

I don't think it should list itself as a dev dependency. Let's just replace file path with fast-querystring. We will push the responsability to the developer to install the dependency while running the benchmark.