Can't run PEX file when a dependency's wheel includes a build tag

[luislew]

We recently encountered a runtime issue with a successfully built PEX file:

(nwpy-ml-clustering_linux-x86-64_cp3.7) app@8bceeea547ee:/srv/nerdwallet/nwpy-ml-clustering$ ./build/nwpy-ml-clustering.pex
Traceback (most recent call last):
  File "/srv/nerdwallet/nwpy-ml-clustering/build/nwpy-ml-clustering.pex/.bootstrap/pex/pex.py", line 391, in execute
  File "/srv/nerdwallet/nwpy-ml-clustering/build/nwpy-ml-clustering.pex/.bootstrap/pex/pex.py", line 89, in _activate
  File "/srv/nerdwallet/nwpy-ml-clustering/build/nwpy-ml-clustering.pex/.bootstrap/pex/environment.py", line 232, in activate
  File "/srv/nerdwallet/nwpy-ml-clustering/build/nwpy-ml-clustering.pex/.bootstrap/pex/environment.py", line 375, in _activate
  File "/srv/nerdwallet/nwpy-ml-clustering/build/nwpy-ml-clustering.pex/.bootstrap/pex/environment.py", line 209, in _update_candidate_distributions
  File "/srv/nerdwallet/nwpy-ml-clustering/build/nwpy-ml-clustering.pex/.bootstrap/pex/environment.py", line 227, in can_add
  File "/srv/nerdwallet/nwpy-ml-clustering/build/nwpy-ml-clustering.pex/.bootstrap/pex/vendor/_vendored/packaging/packaging/tags.py", line 75, in parse_tag
ValueError: too many values to unpack (expected 3)

We tracked this down to the following installed wheel: 'llvmlite-0.29.0-1-cp37-cp37m-manylinux1_x86_64.whl'. The current wheel filename parsing code in environment.py assumes that there is no build tag: https://github.com/pantsbuild/pex/blob/916e61e04634c60d09ee25859c76ba4f27282e51/pex/environment.py#L223

As a result, it includes the build tag with wheel_tags (e.g. 1-cp37-cp37m-manylinux1_x86_64) and parse_tag fails.

Per PEP 427, wheel filenames may include an optional build tag: https://www.python.org/dev/peps/pep-0427/#file-name-convention

---

Proposed fix: pex already vendors wheel, which includes a regex matcher for wheel filenames. Just use the WHEEL_INFO_RE to parse the wheel filename.

Even simpler - split the whole filename, and gather up the trailing 3 tags:

wheel_tags = '-'.join(filename.split('-')[-3:])