perf: make the published benchmark report reproducible and current

[sgerbino]

From the review result:

On the Corosio side make published performance measurements reproducible and current, with sufficient information to understand the comparison.

Source: Boost review result (conditional acceptance), Boost mailing list, September 2026.

The benchmark report's data is from February 2026, measured on the IOCP backend only, and the Test Environment section lists neither CPU, OS build, compiler, Boost/Asio version, corosio commit, nor date. The timer category was removed from the suite but its numbers remain in the summary tables. The repo contains no instructions for running perf/bench and no aggregation scripts. (#343 covers running benchmarks on PRs; this issue covers the published report.)

• Re-run the suite on current develop across backends (IOCP, epoll, io_uring, kqueue)
• Record environment: hardware, OS, compiler, Boost/Asio versions, corosio commit, date
• Drop or clearly segregate the retired timer numbers
• Commit run/aggregate instructions (or scripts) under perf/
• Pin or record the Asio version compared against