test+docs(iace): add handler tests, error-handling tests, JSON export tests, TipTap docs

- Create iace_handler_test.go (22 tests): input validation for InitFromProfile,
  GenerateSingleSection, ExportTechFile, CheckCompleteness, getTenantID,
  CreateProject, ListProjects, Component CRUD handlers
- Add error-handling tests to tech_file_generator_test.go: nil context, nil project,
  empty components/hazards/classifications/evidence, unknown section type,
  all 19 getSystemPrompt types, AI-specific section prompts
- Add JSON export tests to document_export_test.go: valid output, empty project,
  nil project error, special character handling (German text, XML escapes)
- Add iace-hazard-library.md to mkdocs.yml navigation
- Add TipTap Rich-Text-Editor section to iace.md documentation

Total: 181 tests passing (was 165), 0 failures

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

ai-compliance-sdk/internal/iace/tech_file_generator_test.go

@@ -484,6 +484,133 @@ func TestBuildUserPrompt_DeclarationOfConformity(t *testing.T) {
 	}
 }
 
+// ============================================================================
+// Tests: Error handling & edge cases
+// ============================================================================
+
+func TestBuildUserPrompt_NilContext(t *testing.T) {
+	// Should not panic on completely nil context
+	prompt := buildUserPrompt(nil, "risk_assessment_report")
+	if prompt == "" {
+		t.Error("buildUserPrompt should return non-empty string for nil context")
+	}
+	if !strings.Contains(prompt, "Keine Projektdaten") {
+		t.Error("nil context prompt should contain fallback text 'Keine Projektdaten'")
+	}
+}
+
+func TestBuildUserPrompt_NilProject(t *testing.T) {
+	sctx := &SectionGenerationContext{
+		Project: nil,
+	}
+	prompt := buildUserPrompt(sctx, "general_description")
+	if prompt == "" {
+		t.Error("buildUserPrompt should return non-empty string for nil project")
+	}
+	if !strings.Contains(prompt, "Keine Projektdaten") {
+		t.Error("nil project prompt should contain fallback text 'Keine Projektdaten'")
+	}
+}
+
+func TestBuildUserPrompt_EmptyComponents(t *testing.T) {
+	sctx := &SectionGenerationContext{
+		Project:    &Project{MachineName: "Test", Manufacturer: "Corp"},
+		Components: []Component{},
+		Hazards:    nil,
+	}
+	prompt := buildUserPrompt(sctx, "component_list")
+	if !strings.Contains(prompt, "Test") {
+		t.Error("prompt should contain machine name even with empty components")
+	}
+}
+
+func TestBuildUserPrompt_EmptyHazards(t *testing.T) {
+	sctx := &SectionGenerationContext{
+		Project: &Project{MachineName: "Test", Manufacturer: "Corp"},
+		Hazards: []Hazard{},
+	}
+	prompt := buildUserPrompt(sctx, "hazard_log_combined")
+	if prompt == "" {
+		t.Error("prompt should be non-empty even with no hazards")
+	}
+}
+
+func TestBuildUserPrompt_EmptyClassifications(t *testing.T) {
+	sctx := &SectionGenerationContext{
+		Project:         &Project{MachineName: "Test", Manufacturer: "Corp"},
+		Classifications: []RegulatoryClassification{},
+	}
+	prompt := buildUserPrompt(sctx, "classification_report")
+	if prompt == "" {
+		t.Error("prompt should be non-empty even with no classifications")
+	}
+}
+
+func TestBuildUserPrompt_EmptyEvidence(t *testing.T) {
+	sctx := &SectionGenerationContext{
+		Project:  &Project{MachineName: "Test", Manufacturer: "Corp"},
+		Evidence: []Evidence{},
+	}
+	prompt := buildUserPrompt(sctx, "evidence_index")
+	if prompt == "" {
+		t.Error("prompt should be non-empty even with no evidence")
+	}
+}
+
+func TestGetSystemPrompt_UnknownType(t *testing.T) {
+	prompt := getSystemPrompt("totally_unknown_section")
+	if prompt == "" {
+		t.Error("getSystemPrompt should return a fallback for unknown types")
+	}
+	// Fallback should still be a useful CE expert prompt
+	lower := strings.ToLower(prompt)
+	if !strings.Contains(lower, "ce") && !strings.Contains(lower, "experte") && !strings.Contains(lower, "dokumentation") {
+		t.Error("fallback system prompt should reference CE or documentation expertise")
+	}
+}
+
+func TestGetSystemPrompt_AllKnownTypes(t *testing.T) {
+	knownTypes := []string{
+		"risk_assessment_report", "hazard_log_combined", "general_description",
+		"essential_requirements", "design_specifications", "test_reports",
+		"standards_applied", "declaration_of_conformity", "component_list",
+		"classification_report", "mitigation_report", "verification_report",
+		"evidence_index", "instructions_for_use", "monitoring_plan",
+		"ai_intended_purpose", "ai_model_description", "ai_risk_management",
+		"ai_human_oversight",
+	}
+	for _, st := range knownTypes {
+		t.Run(st, func(t *testing.T) {
+			prompt := getSystemPrompt(st)
+			if prompt == "" {
+				t.Errorf("getSystemPrompt(%q) returned empty string", st)
+			}
+		})
+	}
+}
+
+func TestBuildUserPrompt_AIIntendedPurpose(t *testing.T) {
+	sctx := newTestSectionContext()
+	prompt := buildUserPrompt(sctx, "ai_intended_purpose")
+	if prompt == "" {
+		t.Error("prompt should be non-empty for ai_intended_purpose")
+	}
+	if !strings.Contains(prompt, "Robot Arm XY-200") {
+		t.Error("AI intended purpose prompt should contain machine name")
+	}
+}
+
+func TestBuildUserPrompt_AIHumanOversight(t *testing.T) {
+	sctx := newTestSectionContext()
+	prompt := buildUserPrompt(sctx, "ai_human_oversight")
+	if prompt == "" {
+		t.Error("prompt should be non-empty for ai_human_oversight")
+	}
+	if !strings.Contains(prompt, "Robot Arm XY-200") {
+		t.Error("AI human oversight prompt should contain machine name")
+	}
+}
+
 func TestBuildUserPrompt_MultipleHazardAssessments(t *testing.T) {
 	sctx := newTestSectionContext()