[BUG] Binary ops between decimals with high precisions fail

[ChrisJar]

Describe the bug
Performing binary operations between decimal columns with high precisions causes a failure.

Steps/Code to reproduce bug
For example:

import cudf
from cudf.core.dtypes import Decimal64Dtype

s1 = cudf.Series(["1.4", "3.4"]).astype(Decimal64Dtype(10,2))
s2 = cudf.Series(["8.6", "6.3"]).astype(Decimal64Dtype(12,4))
s1 * s2

returns

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-31-b2746bbf79b6> in <module>
      4 s1 = cudf.Series(["1.4", "3.4"]).astype(Decimal64Dtype(10,2))
      5 s2 = cudf.Series(["8.6", "6.3"]).astype(Decimal64Dtype(12,4))
----> 6 s1 * s2

/home/nfs/cjarrett/anaconda3/envs/rapids-gpu-bdb/lib/python3.8/site-packages/cudf/core/frame.py in __mul__(self, other)
   3676 
   3677     def __mul__(self, other):
-> 3678         return self._binaryop(other, "mul")
   3679 
   3680     def __rmul__(self, other):

/home/nfs/cjarrett/anaconda3/envs/rapids-gpu-bdb/lib/python3.8/site-packages/cudf/core/series.py in _binaryop(self, other, fn, fill_value, reflect, can_reindex)
   1341             lhs = self
   1342 
-> 1343         return super()._binaryop(other, fn, fill_value, reflect, lhs)
   1344 
   1345     def add(self, other, fill_value=None, axis=0):

/home/nfs/cjarrett/anaconda3/envs/rapids-gpu-bdb/lib/python3.8/site-packages/cudf/core/frame.py in _binaryop(self, other, fn, fill_value, reflect, lhs, *args, **kwargs)
   3603                     rhs = rhs.fillna(fill_value)
   3604 
-> 3605         outcol = lhs._column.binary_operator(fn, rhs, reflect=reflect)
   3606 
   3607         # Get the appropriate name for output operations involving two objects

/home/nfs/cjarrett/anaconda3/envs/rapids-gpu-bdb/lib/python3.8/site-packages/cudf/core/column/decimal.py in binary_operator(self, op, other, reflect)
     89             )  # precision will be ignored, libcudf has no notion of precision
     90             result = libcudf.binaryop.binaryop(self, other, op, output_type)
---> 91             result.dtype.precision = _binop_precision(
     92                 self.dtype, other.dtype, op
     93             )

/home/nfs/cjarrett/anaconda3/envs/rapids-gpu-bdb/lib/python3.8/site-packages/cudf/core/dtypes.py in precision(self, value)
    279     @precision.setter
    280     def precision(self, value):
--> 281         self._validate(value, self.scale)
    282         self._typ = pa.decimal128(precision=value, scale=self.scale)
    283 

/home/nfs/cjarrett/anaconda3/envs/rapids-gpu-bdb/lib/python3.8/site-packages/cudf/core/dtypes.py in _validate(cls, precision, scale)
    314     def _validate(cls, precision, scale=0):
    315         if precision > Decimal64Dtype.MAX_PRECISION:
--> 316             raise ValueError(
    317                 f"Cannot construct a {cls.__name__}"
    318                 f" with precision > {cls.MAX_PRECISION}"

ValueError: Cannot construct a Decimal64Dtype with precision > 18.0