Created more generic pipeline for text-to-image task

[fgbelidji]

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[arjkesh]

It looks like you'll need to update the corresponding test

sagemaker-huggingface-inference-toolkit/tests/unit/test_diffusers_utils.py

Line 19 in 9923001

from sagemaker_huggingface_inference_toolkit.diffusers_utils import SMAutoPipelineForText2Image

[arjkesh]

One small comment, otherwise lgtm

[arjkesh]

Any particular reason to change the class name? Looks like it is only used here in this file, so should be safe.

[fgbelidji]

I did so as it was related to the import of the class AutoPipelineForText2Image in diffusers not used anymore. Maybe it would make more sense to name it SMDiffusionPipelineForText2Image