refactor lumina2 pipeline tests to the new mixin structure

[akshan-main]

What does this PR do?

Refactors the Lumina2 pipeline tests to the new mixin structure from #14113: a Lumina2PipelineTesterConfig plus TestLumina2Pipeline(PipelineTesterMixin) and TestLumina2PipelineMemory(MemoryTesterMixin), with dummy inputs switched to output_type="pt" and a fresh CPU slice in test_inference.

One note: output_shape is (3, 4, 4), since the dummy one-block VAE decodes the 4x4 latents at scale 1 (the old file never asserted an output shape, so the small size went unnoticed). This unblocks moving tests/lora/test_lora_layers_lumina2.py to the pipeline-level LoRA mixins, which I will follow up with. Follows #14628.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline?
• Did you read our philosophy doc (important for complex PRs)?
• Was this discussed/approved via a GitHub issue or the forum? Discussed on Slack with @sayakpaul
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

Who can review?

@sayakpaul