Support dump_model() / get_dump() for categorical splits

[hcho3]

Currently, it is not possible to use dump_model() or get_dump() with models with categorical splits.

Reproducer:

import pandas as pd
import numpy as np
import xgboost as xgb

rng = np.random.default_rng(seed=0)
x0 = rng.integers(low=0, high=3, size=20)
x1 = rng.integers(low=0, high=5, size=20)
noise = rng.normal(loc=0, scale=0.1, size=20)

df = pd.DataFrame({'x0': x0, 'x1': x1}).astype('category')
y = (x0 * 10 - 20) + (x1 - 2) + noise

params = {'tree_method': 'gpu_hist',
          'predictor': 'gpu_predictor',
          'enable_experimental_json_serialization': True,
          'max_depth': 6,
          'learning_rate': 1.0}

dtrain = xgb.DMatrix(df, label=y, enable_categorical=True)

bst = xgb.train(params, dtrain, num_boost_round=5, evals=[(dtrain, 'train')])
print(bst.get_dump())

This throws the following error:

Traceback (most recent call last):
  File "test.py", line 22, in <module>
    print(bst.get_dump())
  File "/home/phcho/miniconda3/lib/python3.8/site-packages/xgboost/core.py", line 1791, in get_dump
    _check_call(_LIB.XGBoosterDumpModelExWithFeatures(
  File "/home/phcho/miniconda3/lib/python3.8/site-packages/xgboost/core.py", line 188, in _check_call
    raise XGBoostError(py_str(_LIB.XGBGetLastError()))
xgboost.core.XGBoostError: [22:06:05] ../include/xgboost/feature_map.h:85: unknown feature type, use i for indicator and q for quantity
Stack trace:
  [bt] (0) /home/phcho/miniconda3/lib/python3.8/site-packages/xgboost/lib/libxgboost.so(+0x8d264) [0x7f368fa6d264]
  [bt] (1) /home/phcho/miniconda3/lib/python3.8/site-packages/xgboost/lib/libxgboost.so(XGBoosterDumpModelExWithFeatures+0x881) [0x7f368fa65611]
  [bt] (2) /home/phcho/miniconda3/lib/python3.8/lib-dynload/../../libffi.so.7(+0x69ed) [0x7f36f25e09ed]
  [bt] (3) /home/phcho/miniconda3/lib/python3.8/lib-dynload/../../libffi.so.7(+0x6077) [0x7f36f25e0077]
  [bt] (4) /home/phcho/miniconda3/lib/python3.8/lib-dynload/_ctypes.cpython-38-x86_64-linux-gnu.so(+0x1097a) [0x7f36d263297a]
  [bt] (5) /home/phcho/miniconda3/lib/python3.8/lib-dynload/_ctypes.cpython-38-x86_64-linux-gnu.so(+0x110db) [0x7f36d26330db]
  [bt] (6) python(_PyObject_MakeTpCall+0x22f) [0x5602c216350f]
  [bt] (7) python(_PyEval_EvalFrameDefault+0x45d9) [0x5602c21ebd09]
  [bt] (8) python(_PyEval_EvalCodeWithName+0x2d2) [0x5602c21b06a2]

Note to others: The categorical split feature is currently in experimental status.

[trivialfis]

Closing in favor of #6503 .