Handle Anima LoRA alpha keys

[rathodkunj2005]

What

Fixes Anima Comfy-format LoRA conversion when checkpoints include .alpha keys.

The Anima converter already maps diffusion_model.* keys into Diffusers module names, but left .alpha entries in the returned state dict. AnimaLoraLoaderMixin.load_lora_weights() then rejected those keys because they do not contain the lora substring.

This PR folds alpha scaling into the converted LoRA weights and removes the standalone .alpha keys, matching the behavior used by other LoRA converters in this file.

Fixes #14088.

Verification

python -m pytest tests/loaders/test_lora_conversion_utils.py -q
# ..                                                                       [100%]
# 2 passed in 0.05s

python -m ruff check src/diffusers/loaders/lora_conversion_utils.py tests/loaders/test_lora_conversion_utils.py
# All checks passed!