chore(reexecute/c): remove go bench from benchmark

[RodrigoVillar]

Why this should be merged

As part of #4508, this PR removes usage of go bench from the reexecution test. This offers the following benefits:

• Although the reexecution test is housed within a go benchmark, we only run it once, which is not idiomatic as go benchmarks are meant to be iterated over multiple times.
• This sets up the foundation to reintroduce multiple metrics (e.g. block acceptance time, block verification time, etc.) to our dashboard as adding additional metrics is as simple as calling addResult() on the new benchmark tool.

How this works

• Removes usage of *testing.B
• Output of the reexecution statistics file is handled by the test itself rather than ./scripts/benchmark_cchain_range.sh
• Updates GAB (github-action-benchmark) action to use custom benchmarks rather than go bench

How this was tested

CI + ran task reexecute-cchain-range START_BLOCK=1 END_BLOCK=100_000 BENCHMARK_OUTPUT_FILE=<INSERT_CUSTOM_PATH_HERE> locally and got the following:

[
  {
    "name": "BenchmarkReexecuteRange/[1,100000]-Config-default-Runner-dev - mgas/s",
    "value": "254.90228109633716",
    "unit": "mgas/s"
  }
]

Furthermore, results are logged to the terminal as follows:

[12-09|13:06:35.049] INFO c-chain-reexecution c/vm_reexecute.go:637 BenchmarkReexecuteRange/[1,100000]-Config-default-Runner-dev {"result": "262.43886323313035 mgas/s"}

To test that benchmark results display correctly on the GAB dashboard, I created a sample benchmark (https://github.com/RodrigoVillar/sample-reexecution-custom-benchmark/blob/main/main.go) which is a copy-and-paste of the benchmarking code in this PR and plotted some points into my own dashboard. The results of the sample benchmark can be seen below:

Attached is a link to the sample benchmark dashboard: https://rodrigovillar.github.io/sample-reexecution-custom-benchmark/dev/bench/

Need to be documented in RELEASES.md?

No

TODOs after merging

• Prune gh-pages branch of data points which use go bench

[Copilot]

Pull request overview

This PR removes the use of go bench from the C-Chain reexecution test, transitioning it from a Go benchmark to a standalone executable. The change enables more flexible metric reporting and aligns the test with its actual usage pattern (single execution rather than iterative benchmarking).

Key Changes:

• Replaced *testing.B with a custom benchmarkTool for metric collection and reporting
• Changed the package from vm to main and converted the benchmark function to a standalone executable
• Introduced tests.TestContext throughout to manage test lifecycle and cleanup operations

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
tests/reexecute/db.go	Updated CreateBlockChanFromLevelDB signature to use tests.TestContext instead of testing.TB and removed cleanup parameter
tests/reexecute/c/vm_reexecute.go	Converted from benchmark test to main executable; added custom benchmarkTool for metric reporting and JSON output
tests/reexecute/blockexport/main.go	Removed unused cleanup parameter from CreateBlockChanFromLevelDB call
scripts/benchmark_cchain_range.sh	Changed from go test to go run and updated output handling to use JSON file directly
.github/actions/c-chain-reexecution-benchmark/action.yml	Updated benchmark tool from 'go' to 'customBiggerIsBetter' and changed output file extension to .json

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[aaronbuchwald]

Should we remove this and just run the command now?

[RodrigoVillar]

Yup, removed eval here: 955273e

[aaronbuchwald]

How should this be handled if the benchmark output file is not defined?

[RodrigoVillar]

In vm_reexecute.go, we don't write to a file if the benchmark output filepath is empty, but you're right that if a user currently doesn't set BENCHMARK_OUTPUT_FILE, that we'll have:

go run github.com/ava-labs/avalanchego/tests/reexecute/c --benchmark-output-file= --current-state-dir=...`

Changed it so that if a user doesn't define BENCHMARK_OUTPUT_FILE, then --benchmark-output-file isn't set: f682b20