fix(flag-evaluator): return structured scalar evaluation diagnostics

[roncohen]

Summary

• replace direct console.error() calls for numeric, date, and unknown-operator failures with structured evaluation diagnostics
• distinguish invalid context values from invalid targeting values
• report malformed runtime operators as UNKNOWN_OPERATOR
• include relative-date validation that previously failed silently
• avoid including raw context or targeting values in messages
• derive the operator type and runtime allowlist from one source of truth

New diagnostic codes

• INVALID_CONTEXT_VALUE
• INVALID_TARGETING_VALUE
• UNKNOWN_OPERATOR

The feature-event ingest schema accepts arbitrary diagnostic code strings, so these flow through existing evalErrors transport. The web OpenAPI enum should be updated when the new flag-evaluation package is consumed there.

Verification

• cd packages/flag-evaluation && ./node_modules/.bin/vitest run (248 tests)
• tsc --project packages/flag-evaluation/tsconfig.build.json --noEmit
• Node SDK client suite after rebuilding flag-evaluation (132 tests)
• oxlint, oxfmt, and git diff --check

Review

• completed a simplification/refactor pass
• completed a final diff review; no actionable findings

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.