Consolidate and optimize integer power implementations in libcudf

[bdice]

Recently, some issues were identified where std::pow was being used with fixed-point values (#14210, #14233, #14242). There are many places in libcudf where we require an "integer power" operator, because std::pow uses floating-point values and thus gives erroneous results when handling high-precision integral values like decimal128. This thread covers a few key changes that are needed in libcudf:

• Consolidating existing "integer power" operators:

  • cudf/cpp/include/cudf/fixed_point/fixed_point.hpp

    Line 93 in 7825790

    CUDF_HOST_DEVICE inline Rep ipow(T exponent)

  • cudf/cpp/src/binaryop/compiled/operation.cuh

    Line 251 in 7825790

    struct IntPow {

  • Changes from Fix inaccuracy in decimal128 rounding. #14233
  • Changes from Fix inaccurate ceil/floor and inaccurate rescaling casts of fixed-point values. #14242
• Using a lookup table where possible ([FEA] numeric::detail::ipow optimization for Base 10 & 2 #9346), and exponentiation-by-squaring if outside the lookup table's bounds

Related: #10178