Instrument formatter operations to provide visibility into template rendering and response parsing

[psschwei]

Description:
Instrument formatter operations to provide visibility into template rendering and response parsing.

Detailed Requirements:

1. Add spans for formatter operations:
   • formatter.apply - Overall formatter application
   • formatter.render_prompt - Prompt template rendering
   • formatter.parse_response - Response parsing/extraction
2. Add attributes:
   • formatter_type - Class name of formatter
   • template_name - Template identifier if applicable
   • input_length - Input character count
   • output_length - Output character count
3. Handle multiple formatter types in codebase
4. Capture parsing errors as span events

Files to Modify:

• mellea/stdlib/formatters/ - Instrument formatter classes
• mellea/core/formatter.py - Base formatter instrumentation

Acceptance Criteria:

• Formatter spans created during prompt/response processing
• Format type and sizes captured
• Parse errors recorded as span events
• Works with all formatter implementations

[ajbozarth]

Closing as part of the tracing epic (#444) re-plan. This sub-issue predates the plugin/hooks pattern (PR #653) and the GenerationMetadata partition (PR #908); its spec has drifted.

Replacement: part of #1050 — sampling, validator, and formatter spans via plugin hooks (rolled together since the mechanism is identical across all three).

See #444 for the full refreshed plan.