docs: correct 1.6.0 release notes

[sting8k]

Summary

• replace the incomplete Go-only 1.6.0 changelog with coverage of the shipped feature groups
• remove the invalid performance implication and state the current evidence boundary
• keep the source changelog byte-aligned with the corrected public GitHub Release body

Verification

• reviewed by Gizmo with three required edits applied
• verified 5 release archives, 5 SHA-256 assets, and one GitHub build-provenance attestation per archive digest
• git diff --check
• source 1.6.0 section equals the public release-notes body

[greptile-apps]

Greptile Summary

The PR replaces the incomplete Go-focused 1.6.0 changelog with broader coverage of shipped language support, navigation evidence, reliability work, and release artifacts.

• Documents structural owner attribution across supported languages and bounded Go call evidence.
• Describes output compaction, symbol-addressed actions, low-signal advisories, and mixed-language orchestration.
• Adds release-platform, checksum, provenance, and evidence-boundary notes.

Confidence Score: 4/5

The documentation-only PR appears safe to merge after correcting the non-blocking mismatch between its performance-evidence wording and the checked-in benchmark.

The shipped-feature and release-artifact descriptions are supported, but the known-behavior sentence claims additional round-trips against bulk reads even though the repository benchmark measures context pressure against navigation without srcwalk.

Files Needing Attention: CHANGELOG.md

Important Files Changed

Filename	Overview
CHANGELOG.md	Broadens the 1.6.0 release notes substantially, but the revised performance-evidence sentence does not match the comparator, metric, or run counts in the checked-in benchmark.

_{Reviews (1): Last reviewed commit: "docs: correct 1.6.0 release notes" | Re-trigger Greptile}

[greptile-apps]

Performance evidence mismatches benchmark

The revised note says navigation takes additional tool round-trips compared with bulk reads, but the checked-in benchmark compares srcwalk with navigation without srcwalk and measures context pressure rather than round-trips. It also reports reduced context pressure in three of five comparisons, so this wording gives readers an inaccurate account of the recorded evaluation.

Context Used: AGENTS.md (source)