refactor deepfloyd if pipeline tests to the new mixin structure

[akshan-main]

What does this PR do?

Refactors the DeepFloyd IF text-to-image pipeline tests to the new mixin structure from #14113: an IFPipelineTesterConfig plus TestIFPipeline(PipelineTesterMixin) and TestIFPipelineMemory(PipelineOffloadTesterMixin), with the slow class ported to pytest. The dummy components move from the shared IFPipelineTesterMixin in the package __init__.py into the file, so it is self-contained like the other migrated families; the shared mixin gets removed once the last IF file migrates.

Choices verified by running each candidate:

• @skip_mps is kept as the pytest marker it already is: the pipeline outputs all NaN on mps.
• optional_input_params drops latents (the pipelines take no latents argument), mirroring the old required_optional_params subtraction.
• The custom test_save_load_optional_components is ported: the text encoder is optional so the prompt is pre-encoded, where the base test would pass a raw prompt with text_encoder=None.
• The old save_load_local attention-processor override is dropped: the framework version passes without it.
• The float16 save/load override keeps the 1e-1 tolerance for the tiny-random-t5 save/load non-determinism.

test_inference gets a fresh CPU slice. Follows #14577.

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