Google Cloud AIplatform error when saving XGBRegressor model locally

**Description**
When using `aiplatform` to manage experiments, there is an option to [log_model](https://cloud.google.com/python/docs/reference/aiplatform/latest/google.cloud.aiplatform.ExperimentRun#google_cloud_aiplatform_ExperimentRun_log_model). When the XGBRegressor parameter `enable_categorical` is enabled, the default `"model_file": "model.bst"` will fail.

The issue should be caused by the configuration of `_FRAMEWORK_SPECS` defined in `.../lib/python3.10/site-packages/google/cloud/aiplatform/metadata/_models.py`

```Python
_FRAMEWORK_SPECS = {
"sklearn": {
"save_method": _save_sklearn_model,
"load_method": _load_sklearn_model,
"model_file": "model.pkl",
},
"xgboost": {
"save_method": _save_xgboost_model,
"load_method": _load_xgboost_model,
"model_file": "model.bst",
},
"tensorflow": {
"save_method": _save_tensorflow_model,
"load_method": _load_tensorflow_model,
"model_file": "saved_model",
},
}
```

#### Environment details

- OS type and version:
> Windows 11 WSL
> Distributor ID: Ubuntu
> Description: Ubuntu 22.04.2 LTS
> Release: 22.04

- Python version: 3.10.13
- pip version: 24.0
- `google-cloud-aiplatform` version: 1.4.3
- xgboost version: 2.0.3

#### Steps to reproduce
1. Train a XGBoost Regressor model using categorical feature
2. Trying to save the model locally using `.bst` format
3. It will success when set the format to `.json` or `.ubj` as described [here](https://xgboost.readthedocs.io/en/stable/tutorials/saving_model.html#introduction-to-model-io)

#### Code example

```Python
# Necessary imports
import numpy as np
import pandas as pd
import xgboost as xg
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error as MSE

url = 'https://raw.githubusercontent.com/jbrownlee/Datasets/master/housing.csv'
dataset = pd.read_csv(url, header=None)
# Load the data
X, y = dataset.iloc[:, :-1], dataset.iloc[:, -1]

X[8] = X[8].astype('category')

# Splitting
train_X, test_X, train_y, test_y = train_test_split(X, y,
test_size = 0.3, random_state = 42)

# Instantiation
xgb_r = xg.XGBRegressor(enable_categorical=True)

# Fitting the model
xgb_r.fit(train_X, train_y)

import os
import tempfile

model_file = "model.bst"

with tempfile.TemporaryDirectory() as temp_dir:
path = os.path.join(temp_dir, model_file)
print(path)
xgb_r.get_booster().save_model(path)
```

#### Stack trace
```
Stack trace:
[bt] (0) .../lib/python3.10/site-packages/xgboost/lib/libxgboost.so(+0x397223) [0x7fc9a4597223]
[bt] (1) .../lib/python3.10/site-packages/xgboost/lib/libxgboost.so(+0x3985e2) [0x7fc9a45985e2]
[bt] (2) .../lib/python3.10/site-packages/xgboost/lib/libxgboost.so(+0x2ca66e) [0x7fc9a44ca66e]
[bt] (3) .../lib/python3.10/site-packages/xgboost/lib/libxgboost.so(+0x2e67a7) [0x7fc9a44e67a7]
[bt] (4) .../lib/python3.10/site-packages/xgboost/lib/libxgboost.so(XGBoosterSaveModel+0x163) [0x7fc9a433d1c3]
[bt] (5) .../lib/python3.10/lib-dynload/../../libffi.so.8(+0x6a4a) [0x7fca6ae4ba4a]
[bt] (6) .../lib/python3.10/lib-dynload/../../libffi.so.8(+0x5fea) [0x7fca6ae4afea]
[bt] (7) .../lib/python3.10/lib-dynload/_ctypes.cpython-310-x86_64-linux-gnu.so(+0x12461) [0x7fca6ae63461]
[bt] (8) .../lib/python3.10/lib-dynload/_ctypes.cpython-310-x86_64-linux-gnu.so(+0x86eb) [0x7fca6ae596eb]
```

Thanks!