Describe the bug
When a name collision is produced in a schema, by for example referencing to another schema + adding a property to it, an import as is then used. However, the alias is wongly used in the code

To Reproduce
In <my-repo>, dataset.yml + datasets.yml:

---
type: object
title: dataset
properties:
    values:
        type: array
        items:
            type: number

---
type: object
title: datasets
properties:
  datasets:
    type: array
    items:
      allOf:
        - $ref: dataset.yml
      properties:
        name:
          type: string
          default: "dataset"

Used commandline:

$ datamodel-codegen --input <my-repo> --output <my-output-repo> --input-file-type jsonschema

This results in the following datasets.py file:

from __future__ import annotations

from typing import List, Optional

from pydantic import BaseModel

from .dataset import Dataset as Dataset_1


class Dataset(Dataset_1.Dataset):
    name: Optional[str] = 'dataset'


class Datasets(BaseModel):
    datasets: Optional[List[Dataset]] = None

This is obviously wrong as Dataset_1 has no attribute Dataset.

Expected behavior
I would expect the alias to be used properly.
In the example above, I would like to have:

class Dataset(Dataset_1):
    name: Optional[str] = 'dataset'

Version:

• OS: Linux + Windows
• Python version: 3.11
• datamodel-code-generator version: 0.25.2

Additional context
I noticed this behavior is linked with the naming of array items: when a plural (with s at the end) is used for an array field , then the array item with be the same name without the leading s. Other names will result in the Item to be appended.
This smart naming should maybe be dropped to avoid (frequent?) name collision and this alias being wrongly used.