Fix decimal128 fma

[mborland]

Closes: #520

Speeds up multiplication and work around for add step. During the call to normalize in preparation for the add step rounding was occurring which is equivalent just calling operator+.

[codecov]

Codecov Report

Attention: Patch coverage is 89.18919% with 4 lines in your changes are missing coverage. Please review.

Project coverage is 98.7%. Comparing base (94c0413) to head (bdfa306).

Additional details and impacted files

@@            Coverage Diff            @@
##           develop    #523     +/-   ##
=========================================
- Coverage     98.7%   98.7%   -0.0%     
=========================================
  Files          181     182      +1     
  Lines        11507   11500      -7     
=========================================
- Hits         11352   11341     -11     
- Misses         155     159      +4     

Files	Coverage Δ
include/boost/decimal/decimal128.hpp	97.3% <100.0%> (-0.1%)	⬇️
...decimal/detail/cmath/impl/taylor_series_result.hpp	100.0% <ø> (ø)
include/boost/decimal/detail/cmath/tgamma.hpp	100.0% <100.0%> (ø)
...ude/boost/decimal/detail/remove_trailing_zeros.hpp	100.0% <100.0%> (ø)
test/github_issue_519.cpp	100.0% <100.0%> (ø)
test/test_tgamma.cpp	95.8% <20.0%> (-3.3%)	⬇️

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 94c0413...bdfa306. Read the comment docs.