Set up CodSpeed with OIDC authentication and add performance badge

[avalanche-staging]

Summary

This PR updates the existing CodSpeed benchmarks workflow to use OIDC-based authentication and adds the CodSpeed performance badge to the README.

Changes

Workflow updates (.github/workflows/benchmarks.yml):

• Switched from token-based authentication (secrets.CODSPEED_TOKEN) to OpenID Connect (OIDC) by adding id-token: write permission. This is the recommended authentication method and removes the need to manage a secret token.
• Added workflow_dispatch trigger to allow CodSpeed to run backtest performance analysis for generating initial baseline data.

README (README.md):

• Added the CodSpeed performance badge alongside existing project badges.

Existing setup

The repository already has a comprehensive benchmark suite using codspeed-divan-compat (divan) across 25 packages: base64, cksum, cp, cut, dd, df, du, expand, fold, join, ls, mv, nl, numfmt, rm, seq, shuf, sort, split, tsort, unexpand, uniq, wc, factor, and date. All benchmarks run in simulation mode via a matrix strategy.

Next steps

• Once this PR is merged, CodSpeed will automatically run benchmarks on every push and pull request.
• Use the workflow_dispatch trigger to generate initial baseline data if needed.
• Performance results will be available at CodSpeed dashboard.

[avalanche-staging]

Congrats! CodSpeed is installed 🎉

🆕 142 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

• tsort_wide_dag[100000]: 128.8 ms
• tsort_complex_dag[50000]: 84.5 ms
• tsort_linear_chain[1000000]: 1.6 s
• tsort_tree_dag[(10, 3)]: 33.8 ms
• ls_recursive_long_all_wide_tree[(15000, 1500)]: 139.5 ms
• ls_recursive_long_all_balanced_tree[(6, 4, 15)]: 130.2 ms
• ls_recursive_long_all_mixed_tree: 2.8 ms
• ls_recursive_deep_tree[(200, 2)]: 2.2 ms
• ls_recursive_wide_tree[(10000, 1000)]: 51.8 ms
• ls_recursive_mixed_tree: 1.4 ms
• ls_recursive_long_all_deep_tree[(100, 4)]: 3.3 ms
• ls_recursive_balanced_tree[(6, 4, 15)]: 54.6 ms
• seq_custom_separator: 29.3 ms
• seq_with_step: 13.7 ms
• seq_integers: 26.8 ms
• seq_formatted: 97.4 ms
• seq_large_integers: 2.2 ms
• expand_custom_tabstops[50000]: 25.9 ms
• expand_many_lines[100000]: 99.7 ms
• b64_encode_synthetic: 168.1 µs
• ...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

---

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/cp/cp-mv-enotsup-xattr. tests/cp/cp-mv-enotsup-xattr is passing on 'main'. Maybe you have to rebase?
GNU test failed: tests/csplit/csplit-io-err. tests/csplit/csplit-io-err is passing on 'main'. Maybe you have to rebase?
GNU test failed: tests/tail/tail-n0f. tests/tail/tail-n0f is passing on 'main'. Maybe you have to rebase?
Skip an intermittent issue tests/tty/tty-eof (fails in this run but passes in the 'main' branch)
Congrats! The gnu test tests/ls/abmon-align is no longer failing!