Only create a single repaired wheel

[lkollar]

When repairing wheels, the result can be compliant with a lower-level
manylinux spec and in this case auditwheel creates another .whl file for
the lower spec as well. This is unnecessary as the lower spec can always
be used on newer compliant platforms. By only producing a single wheel,
we can probably save some space on PyPI as automated build scripts won't
upload multiple copies of the same wheel.

This also avoids an extra call to repair_wheel which can speed things up
for larger libraries.

Closes: #281

[lkollar]

I haven't added a flag for this because I don't think there is any point in retaining the previous behaviour, but I might be missing something. Anybody see any problems with this?

[codecov]

Codecov Report

Merging #284 (7963262) into master (22c8451) will increase coverage by 0.34%.
The diff coverage is 77.77%.

@@            Coverage Diff             @@
##           master     #284      +/-   ##
==========================================
+ Coverage   89.03%   89.37%   +0.34%     
==========================================
  Files          20       20              
  Lines        1058     1064       +6     
  Branches      226      226              
==========================================
+ Hits          942      951       +9     
+ Misses         69       65       -4     
- Partials       47       48       +1     

Impacted Files	Coverage Δ
auditwheel/main_repair.py	77.35% <77.77%> (-3.50%)	⬇️
auditwheel/main_show.py	64.70% <0.00%> (-5.89%)	⬇️
auditwheel/wheel_abi.py	96.00% <0.00%> (-2.41%)	⬇️
auditwheel/lddtree.py	92.90% <0.00%> (+0.70%)	⬆️
auditwheel/policy/__init__.py	100.00% <0.00%> (+1.85%)	⬆️
auditwheel/wheeltools.py	89.31% <0.00%> (+2.29%)	⬆️
auditwheel/policy/external_references.py	97.61% <0.00%> (+4.76%)	⬆️
auditwheel/main_addtag.py	78.94% <0.00%> (+13.15%)	⬆️

---

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 22c8451...8672e8a. Read the comment docs.

[mayeut]

@lkollar,

I will try to get a little history on the why because I know this has been reported and discussed before.
I fully agree with the fact that only one wheel should be produced however, I do find the information of the policy used to repair the wheel useful and I think there might be a way not to lose it.
I will get back here with my findings over the week-end.

[mayeut]

@lkollar,

I left some comment in #281 (comment)