Commissioning and Quality cpp_double_fp_backend

[ckormanyos]

Here are some open tasks for commissioning and quality enhancement of the proposed cpp_double_fp_backend<T> template

• Make Quickbook docs #165
• Clarify the exact_arithmetic and move additional helper-functions into this struct. #161
• Streamline the ccmath support #164
• Review add/sub/mul/div/sqrt, especially eval_multiply for large LHS #167
• Optimize eval_sqrt. #168
• Ensure that specialized eval_exp and eval_log handle edge-cases (inf, NaN, etc.) #163
• Specialize any needed/wanted Boost.Math templates for the type #176
• Straighten out use of float128_type #169
• Ensure cpp_double_float128 elm-func tests run in CI #172
• Review special values like min/max/epsilon and minimax exponents #175
• Run perf tests stressing simple algebra #178
• Run perf tests stressing simple elem funcs #179

Notes:

• Quickbook docs can be brief, but must be available.
• ccmath support might not be entirely constexpr. It's probably better in some places to simply use existing <cmath> functions (especially those like fma, frexpand ldexp) and change the name of the includes/namespaces.
• For eval_sqrt, see if it is possible to avoid overflow when multiplying the leading limb with the so-called split.

Cc: @sinandredemption and @cosurgi and @jzmaddock