feat(codex): enrich Bash source-read PostToolUse output

[blankanswer]

Summary

• Extend Codex Bash PostToolUse enrichment beyond grep-family output to source-read Bash commands recognized by the existing classifier.
• Normalize Codex cat / head / tail Bash events to the existing Read PostToolUse shape and reuse postRead graph footprint enrichment.
• Preserve the existing grep-family normalization from feat(codex): add Bash PostToolUse graph guidance #211.
• Keep Codex soft-only: only emit hookSpecificOutput.additionalContext; no deny, stop, suppress, or input rewrite.

Design notes

• This is intentionally limited to BashActionReadSource from the existing Bash classifier.
• Reusing runPostToolUse / postRead keeps behavior aligned with existing Claude Code PostToolUse enrichment while remaining inside the Codex dispatcher.
• Claude, Gemini, and Antigravity hook behavior remains unchanged.

Out of scope

• grep-family behavior changes
• sed / awk
• find -name / file-list output
• general Bash output enrichment
• apply_patch
• MCP tool-name matching
• deny / enforcement behavior
• input rewrite
• output suppression
• hook installation / init behavior changes
• Claude / Gemini / Antigravity behavior changes

Tests

• go test ./internal/hooks
• go test ./cmd/gortex
• git diff --check