feat: add tcd sampler

[mashb1t]

closes discussion https://github.com/lllyasviel/Fooocus/discussions/2733

adapted code from https://github.com/comfyanonymous/ComfyUI/pull/3370

TODO: check if virtual scheduler tcd is needed for using sampling_base ModelSamplingDiscreteDistilled or if it's better to use sgm_uniform directly without patching

sgm_uniform

metadata:

{
"prompt": "man in a bar",
"negative_prompt": "",
"prompt_expansion": "man in a bar, epic, dramatic ambient, detailed, relaxed, full focus, expressive, charismatic, amazing detail, intricate, highly coherent, colorful, surreal, incredible light, sharp, symmetry, holy, fine, polished, extremely complex, color, inspiring, beautiful, elegant, decorative, elaborate, gorgeous, awesome, creative, perfect, wonderful, marvelous, pure",
"styles": "['Fooocus V2', 'Fooocus Enhance', 'Fooocus Sharp']",
"performance": "Speed",
"resolution": "(1024, 1024)",
"guidance_scale": 4,
"sharpness": 2,
"adm_guidance": "(1.5, 0.8, 0.3)",
"base_model": "juggernautXL_v8Rundiffusion.safetensors",
"refiner_model": "None",
"refiner_switch": 0.5,
"sampler": "tcd",
"scheduler": "sgm_uniform",
"seed": "2124872720596715030",
"lora_combined_1": "sd_xl_offset_example-lora_1.0.safetensors : 0.1",
"metadata_scheme": "fooocus",
"version": "Fooocus v2.3.1"
}

virtual tcd scheduler (using sgm_uniform, but using ModelSamplingDiscreteDistilled instead of ModelSamplingDiscrete, same as lcm)

metadata:

{
"prompt": "man in a bar",
"negative_prompt": "",
"prompt_expansion": "man in a bar, epic, dramatic ambient, detailed, relaxed, full focus, expressive, charismatic, amazing detail, intricate, highly coherent, colorful, surreal, incredible light, sharp, symmetry, holy, fine, polished, extremely complex, color, inspiring, beautiful, elegant, decorative, elaborate, gorgeous, awesome, creative, perfect, wonderful, marvelous, pure",
"styles": "['Fooocus V2', 'Fooocus Enhance', 'Fooocus Sharp']",
"performance": "Speed",
"resolution": "(1024, 1024)",
"guidance_scale": 4,
"sharpness": 2,
"adm_guidance": "(1.5, 0.8, 0.3)",
"base_model": "juggernautXL_v8Rundiffusion.safetensors",
"refiner_model": "None",
"refiner_switch": 0.5,
"sampler": "tcd",
"scheduler": "tcd",
"seed": "2124872720596715030",
"lora_combined_1": "sd_xl_offset_example-lora_1.0.safetensors : 0.1",
"metadata_scheme": "fooocus",
"version": "Fooocus v2.3.1"
}