v1.0.5 — Code quality, performance, and stability improvements (#26-#40)

[traktore-org]

Summary

Batch release of 15 improvement issues covering performance, stability, usability, accessibility, and code quality. All changes tested on HA-TEST with full validation pass (199 entities, 0 errors).

Changes

Performance (#26, #30)

• Cache forecast source and surplus window between update cycles
• Fix memory leaks and DOM thrashing in dashboard cards (debounced updates, listener cleanup)

Stability (#27, #28, #34, #37, #39, #40)

• Add exponential backoff to EV device retry (10s→320s, max 10 retries)
• Add error recovery for battery protection and load management failures
• Narrow exception handling throughout coordinator (specific types instead of bare except Exception)
• Add storage validation and migration rollback safety
• Entity consistency: availability on number entities, state class last_reset
• Load management: user-set priority preservation, discovery retry, shed list cleanup

Usability (#32, #35, #36, #38)

• Validate entity IDs and sensor values during config flow
• Add notification deduplication and flap suppression (60s stability requirement)
• Replace hardcoded entity IDs with configurable references (consts/core.py)
• Improve dashboard card error states (unavailable entity indicator)

Code Quality (#29, #33)

• Reduce coordinator update method complexity (CC ~25 → ~12)
• Extract shared constants and utilities across dashboard cards (sem-shared.js)

Accessibility (#31)

• Add ARIA labels, role attributes, aria-pressed to dashboard cards

Commits

Linked to individual issues: #26, #27, #28, #29, #30, #31, #32, #33, #34, #35, #36, #37, #38, #39, #40

Test Plan

• Deploy to HA-TEST via deploy-test.sh
• Run validate-sem.sh — all 7 checks passed
• 199 entities, 0 errors in logs
• Coordinator update cycles: 0.001-0.037s (success: True)
• Deploy to HA-PROD

🤖 Generated with Claude Code

[traktore-org]

Implemented and merged in PR #42 (v1.0.5). Tested on HA-TEST with full validation pass.