chore: add codspeed for benchmark

[dqhl76]

Which issue does this PR close?

Closes #4937 .

Rationale for this change

What changes are included in this PR?

A new workflow that runs on every PR and the default main branch.
The criterion dependency is changed to codspeed-criterion-compat.
This should not affect cargo bench

Are there any user-facing changes?

[Xuanwo]

Hi, @dqhl76, token has been created at op://services/codspeed/token.

[codspeed-hq]

CodSpeed Performance Report

Congrats! CodSpeed is installed 🎉

🆕 73 new benchmarks were detected.

You will start to see performance impacts in the reports once the benchmarks are run from your default branch.

Detected benchmarks

• 16.0 MiB (8.5 ms)
• 256 KiB (192.2 µs)
• 4.00 KiB (62.3 µs)
• 4.00 MiB (2.2 ms)
• 16x16.0 MiB (236.9 µs)
• 16x256 KiB (237.2 µs)
• 16x4.00 KiB (234.6 µs)
• 16x4.00 MiB (237.4 µs)
• 1x16.0 MiB (54.5 µs)
• 1x256 KiB (54.7 µs)
• 1x4.00 KiB (54.6 µs)
• 1x4.00 MiB (54.6 µs)
• 2x16.0 MiB (66.9 µs)
• 2x256 KiB (67.2 µs)
• 2x4.00 KiB (66.7 µs)
• 2x4.00 MiB (67 µs)
• 4x16.0 MiB (91.5 µs)
• 4x256 KiB (91.6 µs)
• 4x4.00 KiB (90.2 µs)
• 4x4.00 MiB (91.8 µs)
• 8x16.0 MiB (141.2 µs)
• 8x256 KiB (140 µs)
• 8x4.00 KiB (139.2 µs)
• 8x4.00 MiB (140.7 µs)
• 16.0 MiB (57 µs)
• 256 KiB (56.7 µs)
• 4.00 KiB (56.7 µs)
• 4.00 MiB (56.8 µs)
• 1 (91.6 µs)
• 2 (91.8 µs)
• 4 (91.9 µs)
• 8 (92 µs)
• buffer 256 KiB * 32 advance (660.6 ns)
• buffer 256 KiB * 32 chunk (212.8 ns)
• buffer 256 KiB * 32 truncate (375.3 ns)
• buffer 256 KiB * 4 advance (660.6 ns)
• buffer 256 KiB * 4 chunk (183.6 ns)
• buffer 256 KiB * 4 truncate (375.3 ns)
• buffer 4.00 MiB * 32 advance (660.6 ns)
• buffer 4.00 MiB * 32 chunk (183.6 ns)
• buffer 4.00 MiB * 32 truncate (375.3 ns)
• buffer 4.00 MiB * 4 advance (660.6 ns)
• buffer 4.00 MiB * 4 chunk (183.6 ns)
• buffer 4.00 MiB * 4 truncate (375.3 ns)
• bytes buf 256 KiB * 32 advance (493.3 ns)
• bytes buf 256 KiB * 32 chunk (0 s)
• bytes buf 256 KiB * 32 truncate (552.8 ns)
• bytes buf 256 KiB * 4 advance (493.3 ns)
• bytes buf 256 KiB * 4 chunk (0 s)
• bytes buf 256 KiB * 4 truncate (552.8 ns)
• bytes buf 4.00 MiB * 32 advance (493.3 ns)
• bytes buf 4.00 MiB * 32 chunk (0 s)
• bytes buf 4.00 MiB * 32 truncate (552.8 ns)
• bytes buf 4.00 MiB * 4 advance (493.3 ns)
• bytes buf 4.00 MiB * 4 chunk (0 s)
• bytes buf 4.00 MiB * 4 truncate (552.8 ns)
• 256 KiB * 1000k advance (17.4 ms)
• 256 KiB * 1000k chunk (154.4 ns)
• 256 KiB * 1000k truncate (216.7 ns)
• 256 KiB * 100k advance (1.7 ms)
• 256 KiB * 100k chunk (154.4 ns)
• 256 KiB * 100k truncate (216.7 ns)
• 256 KiB * 10k advance (174.1 µs)
• 256 KiB * 10k chunk (183.6 ns)
• 256 KiB * 10k truncate (216.7 ns)
• 256 KiB * 1k advance (17.9 µs)
• 256 KiB * 1k chunk (154.4 ns)
• 256 KiB * 1k truncate (216.7 ns)
• concurrent 1 (268.8 µs)
• concurrent 16 (324.6 µs)
• concurrent 2 (530.1 µs)
• concurrent 4 (419.1 µs)
• concurrent 8 (361.4 µs)

[Xuanwo]

The benchmark is identified by name, and we can expand it to support various services. We can implement this in future PRs.

[dqhl76]

Hi @Xuanwo, All comments are applied, could you please take a look again? Thanks!

[Xuanwo]

Thank you, @dqhl76, for working on this. Also, thank you, @adriencaccia, for building this great service and helping us with the integration.