Fix iterator version of basic_fields::erase

[inetic]

When used, was causing compile errors.

Not sure whether const_casting here is the best approach,
please let me know if you can think of a better solution and I'll
change it.

[vinniefalco]

Not sure whether const_casting here is the best approach

@pdimov suggests it is correct:

• a precondition of deallocate is that the pointer came from allocate
• allocate gives non-const pointers
• therefore, the element can't have been const to begin with, and the const_cast is safe

[codecov]

Codecov Report

Merging #994 into develop will decrease coverage by 0.01%.
The diff coverage is 100%.

@@             Coverage Diff             @@
##           develop     #994      +/-   ##
===========================================
- Coverage    95.45%   95.43%   -0.02%     
===========================================
  Files          106      106              
  Lines        10601    10614      +13     
===========================================
+ Hits         10119    10130      +11     
- Misses         482      484       +2

Impacted Files	Coverage Δ
include/boost/beast/http/impl/fields.ipp	97.75% <100%> (+0.02%)	⬆️
include/boost/beast/http/fields.hpp	92.3% <0%> (-7.7%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update be14786...bc90b07. Read the comment docs.

[inetic]

Thanks for the info, and happy it's OK. Out of curiosity, would you know what's happening with Travis? The Linux builds took about a day to start and the Mac build has been hanging there for more than two days now.

[pdimov]

Travis has a ~3000 macOS job backlog, https://www.traviscistatus.com/

And the queued macOS jobs reduce the Linux parallelism from 5 to 3, so even Linux slows down.