feat(lora): support PEFT named-adapter LoRAs (e.g. Klein 9B)

[Pfannkuchensack]

Summary

PEFT-format LoRAs with a named adapter encode the adapter name in the weight key (e.g. foo.lora_A.default.weight instead of foo.lora_A.weight). InvokeAI's format detection and conversion only matched the standard PEFT suffix, so these files were classified as Unknown and silently ignored.

Normalize named-adapter keys to the standard PEFT form in both ModelOnDisk.load_state_dict (detection pipeline) and LoRALoader._load_model (conversion pipeline). The pattern is LoRA-specific, so this is a no-op for non-LoRA state dicts. State dicts with multiple distinct adapter names are left untouched to avoid key collisions.

Also widen Flux2-Klein diffusers detection to accept keys without a transformer. or base_model.model. prefix — some trainers (Modelscope / MuseAI Klein 9B finetunes) save at the top level.

Related Issues / Discussions

#9021 and #9249

QA Instructions

Tested with a Klein 9B LoRA from Modelscope (klein-9b-turn2real_v1.5_ep9.safetensors) that uses the PEFT named-adapter format and no key prefix.

1. Drop a PEFT named-adapter LoRA into your models directory (any base — SD1, SDXL, FLUX, FLUX.2 Klein, Z-Image, Qwen Image).
   • Confirmation that a file is affected: its keys look like …lora_A.default.weight / …lora_B.default.weight rather than …lora_A.weight / …lora_B.weight.
2. Trigger a model scan / import.
3. Before this PR: the file is classified as Unknown (no base, no type, no format).
   After this PR: the file is recognized with the correct base, type=LoRA, format=LyCORIS, and (for Flux2) the correct variant (Klein4B / Klein9B).
4. Use the LoRA in a generation and verify it actually patches the model (no shape errors at load time, visible effect on the output).

Automated coverage:

• tests/backend/patches/lora_conversions/test_peft_adapter_utils.py — 6 unit tests for the normalization helper (no-op when not needed, strips default, strips custom names, leaves multi-adapter dicts alone, preserves non-LoRA / integer keys).
• All existing tests/backend/patches/lora_conversions/ tests (108) and tests/backend/model_manager/configs/ tests (41) still pass.

Merge Plan

Standard merge — no schema/migration changes.

Checklist

• The PR has a short but descriptive title, suitable for a changelog
• Tests added / updated (if applicable)
• ❗Changes to a redux slice have a corresponding migration
• Documentation added / updated (if applicable)
• Updated What's New copy (if doing a release after this PR)