#23161: fix(imessage): parse permission denied errors from imsg rpc plaintext output

Fixes #23133 ## Problem When imsg rpc encounters macOS permission issues accessing chat.db, it emits plaintext error messages like: ``` permissionDenied(path: "/Users/.../chat.db", underlying: authorization denied (code: 23)) ``` These lines fail JSON parsing in `IMessageRpcClient.handleLine`, causing: - Generic 'failed to parse' logs with no actionable guidance - Channel crashes without clear diagnosis - Users unable to identify the permission issue ## Solution 1. Add `parseIMessageRpcPlaintextError()` helper to detect permission errors 2. When JSON parse fails, check if line matches `permissionDenied` pattern 3. Extract path and underlying error details 4. Construct actionable error with Full Disk Access remediation 5. Call `failAll()` to propagate error with proper context 6. Mark permission errors as fatal in `probeIMessage` for fast-fail behavior ## Testing - ✅ Added `client.parse-plaintext-error.test.ts` with permissionDenied cases - ✅ Updated `probe.test.ts` to verify fatal flag on permission errors - ✅ All existing iMessage tests pass - ✅ Full quality gate passes (build, tsgo, lint, test) ## Impact Users hitting permission issues now see clear guidance: ``` imsg permission denied for /path/to/chat.db (authorization denied); grant Full Disk Access to the process running OpenClaw/imsg and restart ``` This matches the user's reported issue where `permissionDenied` output was being logged as 'failed to parse...Unexpected token' with no hint about the actual macOS permission problem. --- **AI-assisted:** Implementation and tests generated with Claude Sonnet 4.5 via OpenClaw autonomous contributor workflow. <!-- greptile_comment --> <h3>Greptile Summary</h3> Adds plaintext error parsing to detect macOS permission errors from `imsg rpc` output. When JSON parsing fails, the code now checks for `permissionDenied` patterns and constructs actionable errors with Full Disk Access remediation guidance. The `fatal` flag propagates through probe failures to enable fast-fail behavior. <h3>Confidence Score: 5/5</h3> - This PR is safe to merge with minimal risk - The changes are well-scoped to error handling, properly tested with unit tests covering both quoted and unquoted path scenarios, and follow existing patterns in the codebase. The regex-based parsing is simple and defensive (returns null for non-matches), and the fatal flag integration is consistent with existing error handling logic. - No files require special attention _{Last reviewed commit: 460b83e} <!-- greptile_other_comments_section --> <!-- /greptile_comment -->