fix: disable benchmark runs for feature branches

[MANASAB805]

Description

This PR modifies the PR benchmarks workflow to skip running benchmarks for feature branches, as they can be time-consuming and may clutter the benchmark history. Benchmarks will now only run for:

• Release branches (release/*)
• Hotfix branches (hotfix/*)
• Manual triggers via workflow_dispatch

Changes

• Added workflow_dispatch trigger for manual benchmark runs
• Modified branch conditions to skip feature branches
• Added documentation for the workflow conditions

Testing

• Verified that the workflow conditions work as expected in a test repository
• Confirmed that the workflow can be manually triggered

Related Issues

Closes #463

[jonasstrehle]

This does not solve the issue though, can you double check?

[MANASAB805]

@jonasstrehle I've updated the PR to correctly exclude feature branches. The workflow is now working as expected:

Changes made:

• Changed the condition from only allowing specific branch types (release/*, hotfix/*) to excluding feature/* branches using !startsWith(github.head_ref, 'feature/')
• Fixed the workflow_dispatch syntax error
• Benchmarks will now run for all branches except feature/* branches, and can be manually triggered when needed

The workflow will now:

• ✅ Skip benchmarks for feature/* branches
• ✅ Run benchmarks for release/*, hotfix/*, fix/*, chore/*, and other branches
• ✅ Allow manual triggering via workflow_dispatch

[jonasstrehle]

please remove the dry run here, we don't have any benefit in doing this, as we already perform the dry run in the release/xx branch as required action to be passed before we can merge.

[jonasstrehle]

Could you take a look at the two comments and also rebase this branch to the release/0.0.7 please?
If fixed, I'll approve and merge. Thanks for the contributions @MANASAB805 :)

[MANASAB805]

@jonasstrehle I've addressed all your feedback:

1. Removed Redundant Dry-Run Step:

   • Removed the cargo publish --dry-run step from publish.yml as requested, since it was redundant with the release branch checks.

2. Workflow Inputs:

   • Kept the workflow_dispatch inputs in pr_benchmarks.yml as they provide useful context for manual runs, but they remain optional.

3. Rebased on release/0.0.7:

   • Successfully rebased the branch on the latest release/0.0.7.

The PR is now ready for your review. All checks have passed, and the changes are up to date with the target branch.

Changes Made:

• Removed redundant dry-run step from publish workflow
• Maintained optional workflow inputs for better traceability
• Ensured proper branch conditions for benchmark execution

Let me know if you need any further adjustments!