[reach2sayan]

Fixes #67.

rounding_control<long double>::to_int called rint, the double function, in three backends, so a long double was rounded to 53 bits before it was rounded to an integer. Above 2^53, int_down and int_up returned wrong values, sometimes on the wrong side of the argument, and fmod, which uses int_down, returned intervals that do not contain the true result: fmod(2^60 + 1, 1) gave [1, 1].

Not affected: 32-bit x86 (x87 frndint), and targets where long double is double (MSVC, 32-bit ARM).

Test: test_long_double() in test/fmod.cpp checks that fmod(2^60 + 1, 1) contains 0. It is skipped when long double has fewer than 62 significant bits.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.85%. Comparing base (3a640df) to head (4c46d2c).

Additional details and impacted files

@@             Coverage Diff             @@
##           develop      #71      +/-   ##
===========================================
+ Coverage    82.79%   82.85%   +0.06%     
===========================================
  Files           23       23              
  Lines         1017     1021       +4     
  Branches       412      414       +2     
===========================================
+ Hits           842      846       +4     
  Misses         167      167              
  Partials         8        8              

Files with missing lines	Coverage Δ
...umeric/interval/detail/c99sub_rounding_control.hpp	100.00% <100.00%> (ø)

... and 2 files with indirect coverage changes

---

Continue to review full report in Codecov by Harness.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3a640df...4c46d2c. Read the comment docs.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.