[python] disabled split value histogram for categorical features

[StrikerRUS]

Replace TypeError for categorical features with more understandable LightGBMError.

import lightgbm as lgb
from sklearn.datasets import load_boston

data = lgb.Dataset(*load_boston(True), categorical_feature=[2])
booster = lgb.train({'verbose': -1}, data, num_boost_round=20)
booster.get_split_value_histogram(2)

TypeError                                 Traceback (most recent call last)
<ipython-input-4-46f3fc73cd0a> in <module>
----> 1 booster.get_split_value_histogram(2)

C:\Program Files\Anaconda3\lib\site-packages\lightgbm\basic.py in get_split_value_histogram(self, feature, bins, xgboost_style)
   2480             bins = max(min(n_unique, bins) if bins is not None else n_unique, 1)
   2481         print(values)
-> 2482         hist, bin_edges = np.histogram(values, bins=bins)
   2483         if xgboost_style:
   2484             ret = np.column_stack((bin_edges[1:], hist))

C:\Program Files\Anaconda3\lib\site-packages\numpy\lib\histograms.py in histogram(a, bins, range, normed, weights, density)
    700     a, weights = _ravel_and_check_weights(a, weights)
    701 
--> 702     bin_edges, uniform_bins = _get_bin_edges(a, bins, range, weights)
    703 
    704     # Histogram is an integer or a float array depending on the weights.

C:\Program Files\Anaconda3\lib\site-packages\numpy\lib\histograms.py in _get_bin_edges(a, bins, range, weights)
    353             raise ValueError('`bins` must be positive, when an integer')
    354 
--> 355         first_edge, last_edge = _get_outer_edges(a, range)
    356 
    357     elif np.ndim(bins) == 1:

C:\Program Files\Anaconda3\lib\site-packages\numpy\lib\histograms.py in _get_outer_edges(a, range)
    248         first_edge, last_edge = 0, 1
    249     else:
--> 250         first_edge, last_edge = a.min(), a.max()
    251         if not (np.isfinite(first_edge) and np.isfinite(last_edge)):
    252             raise ValueError(

C:\Program Files\Anaconda3\lib\site-packages\numpy\core\_methods.py in _amin(a, axis, out, keepdims, initial)
     30 def _amin(a, axis=None, out=None, keepdims=False,
     31           initial=_NoValue):
---> 32     return umr_minimum(a, axis, None, out, keepdims, initial)
     33 
     34 def _sum(a, axis=None, dtype=None, out=None, keepdims=False,

TypeError: cannot perform reduce with flexible type

values here
https://github.com/Microsoft/LightGBM/blob/8d6666e0ffb7165096753dc392044af18ef2eae6/python-package/lightgbm/basic.py#L2481
values = ['2||3||6||10||18', '2||3||6||10||18', '2||4||6||10||18', '2||3||6||10||18', '1||4||5||6', '1||2||4||10||21', '1||4||5||6', '10||18||19||21', '3||6||7||8||10||18||21', '3||5||6||9', '1||2||3||7||10||21', '1||2||3||8||10||21', '5||6||10||18', '5||6||9||18', '6||7||9||10||18', '2||4||6||7||9||18', '1||2||3||8||21']