[5.x] Add `bench.all` to group benchmarks without comparing them

Clear and concise description of the problem

With the recent benchmarking API overhaul, I'd love for there to be a new bench.all API that works similar to bench.compare, but with some notable changes:

1. The results are sorted by argument order and not by speed
2. Each run will get the same options, and not do more iterations just because it's faster than others in the group.

Suggested solution

A new bench.all API, similar to bench.compare.

Not sure if tinybench/vitest already does this, but it might be good to have a default of "interleaved" runs, and only do "sequential" benchmarking when configured. This would allow for more real-life use cases when benchmarking I feel.

Alternative

No response

Additional context

Currently, running benchmarks individually will show each benchmark as its own little table, making the results very hard to read/parse.

This issue was opened after a Discord discussion: https://discord.com/channels/917386801235247114/917386801235247117/1545405463972282480

Validations

• Follow our Code of Conduct
• Read the Contributing Guidelines.
• Read the docs.
• Check that there isn't already an issue that requests the same feature to avoid creating a duplicate.

Start by tracing the existing bench.compare API and review the benchmarking behavior discussed for tinybench/vitest. Define how bench.all should preserve argument order, share options and present grouped results, including whether interleaved runs are the default. Done means the new API is covered by tests for ordering, iteration options and readable grouped output.