Add Chroma modular pipeline (text2image)

[charchit7]

What does this PR do?

Fixes #13295. Adds a ModularPipeline for Chroma (t2i only):

• T5-only text encoder block with Chroma's padding-token attention masking
• guider-based CFG denoise loop (ClassifierFreeGuidance)
• reuses Flux-style packed-latent prep, mu-shift timesteps, RoPE ids, decode

Ran the self-review:

• Verdict: READY

Before submitting

• Did you use an AI agent (Claude Code, Codex, Cursor, etc.) to help with this PR? If so:
  • Did you read the Coding with AI agents guide?
  • Did you run the self-review skill on the diff?
  • Did you share the final self-review notes in the PR description or a comment?
• 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? Please add a link to it if that's the case.

Who can review?

@yiyixuxu @dg845 and @asomoza

[charchit7]

config:

import os
os.environ["HF_HOME"] = "/home/jupyter/images/hf_cache"

import numpy as np
import torch
from PIL import Image

from diffusers import ChromaPipeline
from diffusers.guiders import ClassifierFreeGuidance
from diffusers.modular_pipelines import ChromaAutoBlocks

MODEL_ID = "lodestones/Chroma1-HD"

PROMPT = (
    "A painting of a squirrel eating a burger, highly detailed, vibrant colors, "
    "professional art, sharp focus, cinematic lighting"
)

NEGATIVE_PROMPT = (
    "low quality, ugly, unfinished, out of focus, deformed, "
    "disfigure, blurry, smudged, restricted palette, flat colors"
)

HEIGHT = WIDTH = 1024
NUM_STEPS = 40
GUIDANCE_SCALE = 3.0
MAX_SEQ_LEN = 512

OUTPUT_DIR = "/home/jupyter/images/chroma_output"
os.makedirs(OUTPUT_DIR, exist_ok=True)

DTYPE = torch.bfloat16
DEVICE = "cuda"
SEED = 42

Modular output :

standard output: