Prelim fix issue586 (if none better found)

[ckormanyos]

The purpose of this PR is to provide a minimalistic fix to #586.

In that issue, we see a phenomenon that may require conversion to number-wrap or non-wrapped type detected in overloads of trunc and lltrunc.

If no better fix is found in this releasce cycle, then this does the trick.

Cc: @jzmaddock John take a look if you find better...

Cc: @mborland Standalone with/without the presence of Math.

Cc: @radj307

[ckormanyos]

It's green-ish with CodeCov upload error (re-running) and 1 timeout on drone.

Hi @jzmaddock do you think the potential fix in this PR is sufficient?

Cc: @mborland

[jzmaddock]

I'll take a look shortly.

[jzmaddock]

That's fine, a simpler fix is:

    if (arg > 0) 
       return floor(arg);
    return ceil(arg);

which doesn't then require a temporary.

[ckormanyos]

That's fine, a simpler fix is:

Thank you John. It is now cycling in CI.

[codecov]

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (878138a) 94.1% compared to head (29ca08c) 84.6%.

Additional details and impacted files

@@            Coverage Diff             @@
##           develop    #587      +/-   ##
==========================================
- Coverage     94.1%   84.6%    -9.5%     
==========================================
  Files          273     149     -124     
  Lines        28523   16332   -12191     
==========================================
- Hits         26833   13805   -13028     
- Misses        1690    2527     +837     

Files	Coverage Δ
...de/boost/multiprecision/detail/functions/trunc.hpp	84.7% <66.7%> (-6.2%)	⬇️

... and 159 files with indirect coverage changes

---

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 878138a...29ca08c. Read the comment docs.

[ckormanyos]

Hi Matt (@mborland) in Multiprecision, uploading Coverage from suite number 1 takes too long and times out all the time, or a lot of the time. We need to eventually look into this.

I can help when we both have a time slot for this...

Cc: @jzmaddock

[mborland]

In math every test suite gets its own coverage run because just doing suite 1 and suite 2 also had issues with timeouts. Porting that logic is likely the easiest way to solve this.

[ckormanyos]

every test suite gets its own coverage run

I appreciate that. Many thanks @mborland