Fix: telemetry decorator wiring

[kddejong]

Improve telemetry accuracy and troubleshooting capabilities

This PR improves telemetry metrics to better distinguish between service availability issues and infrastructure/developer
issues, while adding rich error attributes for troubleshooting.

Changes

Telemetry Decorator Wiring

• Fixed DiagnosticCoordinator and PyodideWorkerManager to use @telemetry() decorator instead of manual ScopedTelemetry instantiation
• Ensures consistent telemetry initialization across all services

Availability Metrics Accuracy

• Added errorType classification to lint.error and validate.error metrics
• Infrastructure errors (worker crashes, uninitialized worker, mount failures) are counted with distinct errorType values
• Parse errors excluded from Guard validate.error (tracked separately as parser.error)
• Infrastructure issues also tracked separately: worker.crash, init.fault, mount.fault
• Enables CloudWatch dashboards to filter on errorType attribute to distinguish:
  • cfn-lint availability: lint.error where errorType NOT IN ('WorkerCrash', 'WorkerNotInitialized', 'MountError')
  • Infrastructure health: worker.crash, init.fault, mount.fault, or filter lint.error by infrastructure errorTypes
  • User impact: All lint.error regardless of errorType

Mount Error Handling

• Added MountError class for proper error classification
• Non-GitSync files automatically fall back to standalone linting when mount fails
• Mount failures tracked as mount.fault with errorType attribute
• Improves resilience while maintaining visibility into mount issues

Error Attributes for Troubleshooting

• All error metrics now use telemetry.error() with captureErrorAttributes: true
• Captures secure metadata without exposing user data:
  • error.type: Error class name
  • error.origin: Where it originated
  • errorType: Semantic classification (WorkerCrash, Timeout, PythonError, etc.)
  • Stack location: File and line number
• Applies to: init.fault, lint.error, mount.fault, validate.error, parser.error, wasm.error, memory.threshold.exceeded,
  rules.load.error

Testing

• Updated test expectations for MountError message format
• All existing tests pass

Metrics Impact

This PR maintains backward compatibility for metric names but adds attributes that enable better filtering and analysis
in CloudWatch dashboards.