[PullsarDev]

Summary

Documentation-only follow-up to the published v0.2.0 release.

• Replace an unreachable pre-merge revision reference and pre-release "candidate" labels in the benchmark report with v0.2.0 implementation terminology.
• Explicitly disclose that the recorded timings came from the final pre-release Step implementation, not a newly benchmarked release-tag build. The published tag has identical src/step.cpp, include/interest/attention.hpp, and benchmark harness blobs; version/documentation metadata changed after measurement.
• Explain that the attention policy does not restore a previous hysteresis-held target after a transient event. Once dwell protection ends, a decayed event target may produce Clear when no challenger reaches the acquisition threshold.
• Simplify the README introduction, linking to the existing v0.2.0 release notes for field-layout and upgrade details.

Scope and verification

Changed files: README.md, docs/benchmarks.md only. No production code, tests, numerical behavior, version definitions, tag or release changes.

Cross-checked the documented Clear condition against ApplyTemporalPolicy; compared upstream Git blob hashes for the Step implementation and benchmark harness between the measured pre-release revision and the published v0.2.0 tag. Commit author/committer metadata uses GitHub noreply.

No build or benchmark was rerun for these Markdown-only changes. The published release's pre-existing CI remains unchanged.

Draft for review. Do not merge automatically.