fix musl float arithmetic and move it into sub-library

[larskuhtz]

On some systems the c implementation of transcendental float arithmetic could result in wrong results due to conflicts with symbols in the system libm.

This PR

• simplifies the implementation of transcendental float arithmetic by removing redundant code and making all symbols that aren't used directly by Haskell FFI static, and
• moves the implementation into a sub-library in order to make debugging and testing easier.

[imalsogreg]

Is this PR obsoleted by https://github.com/kadena-io/pact/pull/1184 ?

[jwiegley]

Is this PR obsoleted by #1184 ?

I would call this PR an improvement upon it. Except that now, this PR is only doing a pure code reorg, and no longer attempting to fix an issue.