Bucket support

### Feature Idea

Hi guys,

I have a bucket with 500gb of various AI models, every day we start 10 GPU servers and I mount that bucket via s3fs to models/ directory. However it performs quite slow.

Is there already a native support for s3 compatible API, so instead of a filesystem Comfy UI uses API to access models? or something similar?

### Existing Solutions

actually i didn't think of searching a custom node, I'll update the request if I find one.

### Other

my example colab code:

```
USE_R2_BUCKET = True #@param {type:"boolean"}
if USE_R2_BUCKET:
r2_id = userdata.get('r2_id')
r2_key = userdata.get('r2_key')
passwd_file_path = '/content/s3fs.passwd'
cache_dir = '/content/r2_cache'
!mkdir -p '/content/r2_cache'
with open(passwd_file_path, 'w') as f:
f.write(f"{r2_id}:{r2_key}\n")
!chmod 600 {passwd_file_path}
if is_mounted('/content/r2_bucket'):
!umount /content/r2_bucket
!s3fs colab-models /content/r2_bucket -o parallel_count='16' -o multipart_size="128" -o use_cache={cache_dir} -o passwd_file={passwd_file_path} -o url=https://mysecretpath.r2.cloudflarestorage.com -o use_path_request_style,url=https://mysecretpath.r2.cloudflarestorage.com -d
!rm '/content/s3fs.passwd'
```