feat(car): raise the maximum supported cars from 4 to 8

[springfall2008]

Summary

Follow-up to #4533. Some users have more than 4 EVs/chargers on their account (e.g. multiple Octopus Intelligent devices), which the previous 4-car ceiling rejected by clamping num_cars down to 4. This raises the supported maximum to 8.

Three things had to move together:

• PREDBAT_MAX_CARS in const.py: 4 → 8 (single source of truth, already used by the fetch_config_options() clamp added in fix(car): clamp num_cars to supported maximum, exclude suspended Octopus devices #4533)
• The C++ kernel's PK_MAX_CARS array size in prediction_kernel.cpp, with PK_PARITY_REVISION/KERNEL_PARITY_REVISION bumped 5 → 6 so a stale (still-4-car) binary is safely rejected at load time and falls back to the Python engine instead of corrupting the ctypes struct layout
• New car_charging_rate_4..7 and car_charging_manual_soc(_kwh)_4..7 entries in config.py's CONFIG_ITEMS so cars 4-7 get real HA entities to configure (mirrors the existing 0-3 pattern exactly)

Bug caught during development

The Python-side ctypes.Structure fields (car_charging_limit, car_charging_soc, car_soc_next) hardcoded array size 4 directly instead of referencing KERNEL_MAX_CARS. Bumping only the constant caused a segfault in the kernel parity test - the C++ side wrote 8-car arrays into a struct Python had only sized for 4. Fixed by making those three fields reference KERNEL_MAX_CARS like everything else already did.

All 6 cross-compiled kernel binaries (prediction_kernel_lib_*.so) are rebuilt and committed directly in this PR (via build_kernel_cross.sh with zig), so CI's binary-verification step should pass on the first run rather than depending on the auto-fix job's follow-up commit.

Test plan

• ./run_all --test kernel_parity - passes with a genuinely rebuilt kernel loaded (not the Python fallback), 450-configuration random sweep included
• ./run_all --test fetch_config_options --test octopus_misc - new/updated tests pass, including a real (non-mocked) config_index resolution of all 8 cars
• python3 verify_kernel_binary.py prediction_kernel_lib_darwin_arm64.so - the cross-built binary (same compiler/flags as the Linux targets, different triple) loads and passes the full parity suite locally
• ./run_all --quick - full suite passes
• ./run_pre_commit - passes

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Raises Predbat’s supported EV/charger count from 4 to 8 by updating the single source-of-truth maximum, extending Home Assistant config entities for the additional cars, and expanding the C++ prediction kernel’s struct/array sizes with a parity revision bump to safely reject stale binaries.

Changes:

• Increased PREDBAT_MAX_CARS to 8 and updated tests to assert the new clamp behavior.
• Extended HA configuration (CONFIG_ITEMS) to include car_charging_rate_4..7 and car_charging_manual_soc(_kwh)_4..7.
• Updated prediction kernel parity revision and car array sizing on both Python (ctypes.Structure) and C++ sides.

Reviewed changes

Copilot reviewed 7 out of 13 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
apps/predbat/tests/test_octopus_misc.py	Updates test commentary around device counting behavior.
apps/predbat/tests/test_infra.py	Extends mocked default config with car_charging_rate_4..7.
apps/predbat/tests/test_fetch_config_options.py	Updates clamp test for 8 cars and adds a regression test ensuring all 8 car rates resolve via real config_index.
apps/predbat/prediction_kernel.py	Bumps parity revision and fixes ctypes car-array fields to size by KERNEL_MAX_CARS.
apps/predbat/prediction_kernel.cpp	Bumps parity revision and increases PK_MAX_CARS to 8.
apps/predbat/const.py	Raises PREDBAT_MAX_CARS to 8.
apps/predbat/config.py	Adds HA config items for car indices 4–7 (rate + manual SOC controls).

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.