fix(security): track identifier lines linearly

[alxxjohn]

Motivation

• The identifiers-in-source routine computed each identifier's line by counting newlines from the file start for every match, producing O(n^2) work on identifier-dense inputs and enabling CI/scan denial-of-service.
• The goal is to preserve existing naming checks while eliminating repeated scanning of growing prefixes so scans remain available for untrusted repositories.

Description

• Rewrote identifiersInSource in internal/codeguard/checks/quality/quality_precision_workstreams_cd.go to maintain a running line counter and only strings.Count the non-overlapping slice between consecutive matches, making line computation linear-time.
• Added internal/codeguard/checks/quality/quality_precision_workstreams_cd_test.go with TestIdentifiersInSourceTracksLines to assert correct line numbers across same-line identifiers, blank lines, and subsequent line transitions.
• Preserved identifier names and output shape so downstream naming checks (e.g., roleSuffixOveruseFinding, crossLayerInconsistencyFinding) continue to operate without behavior changes.

Testing

• Formatted the modified files with gofmt and ran go test ./internal/codeguard/checks/quality ./tests/checks, and the test suites completed successfully.
• Ran go test specifically for the internal/codeguard/checks/quality package and observed the new unit test pass.
• Attempted szr-wrapped test command but szr is not installed in this environment, and equivalent direct go test invocations were used instead and passed.

---

Codex Task