Precision fixes for some code dealing with ArbField/AcbField

[fingolfin]

Fixes #1580 reported by @fredrik-johansson which was introduced by PR #1367 resp. commit eefcf5b.

I am not quite sure if the test I added is "right" to verify we get the desired precision, if someone can suggest a better methods, I'd be obliged. (I don't quite understand why radius(y) suggest a precision of about 1e-113 while the printing says ... +/- 1.79e-101 -- I wanted to check the latter, which nicely patches $\log_{10}(2^{333})$ but couldn't find a good way).

I also fixed a few other things in the same PR but did not add any tests for them.

Moreover, a few of the changed functions should probably get variants for RealField and ComplexField..

[thofma]

In Hecke we wrapped

function radiuslttwopower(x::ArbFieldElem, e::Int)
  GC.@preserve x begin
    t = ccall((:arb_rad_ptr, libarb), Ptr{Nemo.mag_struct}, (Ref{ArbFieldElem}, ), x)
    b = ccall((:mag_cmp_2exp_si, libarb), Cint,
            (Ptr{Nemo.mag_struct}, Int), t, e) <= 0
  end
  return b
end

which tests whether the radius is less than $2^e$.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.75%. Comparing base (f033978) to head (9602602).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1707   +/-   ##
=======================================
  Coverage   84.74%   84.75%           
=======================================
  Files          94       94           
  Lines       37205    37204    -1     
=======================================
+ Hits        31530    31531    +1     
+ Misses       5675     5673    -2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[fingolfin]

Yay, all green