refactor: migrate workflow system from 6-field nested to unified flat task schema

- Schema: add shared_context to plan-overview-base-schema, add pre_analysis/artifacts/inherited
  and polymorphic implementation (string|object with tdd_phase) to task-schema
- Producer: action-planning-agent outputs flat fields (description, depends_on, focus_paths,
  convergence.criteria, files, implementation, pre_analysis) + plan.json generation
- Orchestrator: plan.md/tdd-plan.md validate plan.json, task-generate-agent/tdd output dual-layer
- Consumer: code-developer/tdd-developer/test-fix-agent/universal-executor read flat fields
- Execute/review: read plan.json for execution strategy, use flat field paths
- Remove all migration notes referencing old field names

.claude/agents/test-fix-agent.md

@@ -59,7 +59,7 @@ jq --arg ts "$(date -Iseconds)" '.status="in_progress" | .status_history += [{"f
 ### Flow Control Execution
 When task JSON contains `flow_control` field, execute preparation and implementation steps systematically.
 
-**Pre-Analysis Steps** (`flow_control.pre_analysis`):
+**Pre-Analysis Steps** (`pre_analysis`):
 1. **Sequential Processing**: Execute steps in order, accumulating context
 2. **Variable Substitution**: Use `[variable_name]` to reference previous outputs
 3. **Error Handling**: Follow step-specific strategies (`skip_optional`, `fail`, `retry_once`)
@@ -72,7 +72,7 @@ When task JSON contains `flow_control` field, execute preparation and implementa
 "Glob(pattern)"         → Glob tool: Glob(pattern=pattern)
 ```
 
-**Implementation Approach** (`flow_control.implementation_approach`):
+**Implementation Approach** (`implementation`):
 When task JSON contains implementation_approach array:
 1. **Sequential Execution**: Process steps in order, respecting `depends_on` dependencies
 2. **Dependency Resolution**: Wait for all steps listed in `depends_on` before starting
@@ -162,7 +162,7 @@ run_test_layer "L1-unit" "$UNIT_CMD"
 
 ### 3. Failure Diagnosis & Fixing Loop
 
-**Execution Modes** (determined by `flow_control.implementation_approach`):
+**Execution Modes** (determined by `implementation`):
 
 **A. Agent Mode (Default, no `command` field in steps)**:
 ```