Remove shellify

[nth10sd]

pathlib2 from PyPI (or pathlib via the Python 3.5+ stdlib) and shellescape from PyPI are the new goodness to use. This is because the custom functions getAbsPathForAdjacentFile, normExpUserPath and shellify do not have tests.

This PR starts us off the route for using pathlib2/pathlib in our tests first, and completely removes shellify. Fixes #183.

[nth10sd]

Due to test failures, I will have to split off pathlib2 additions into a separate PR, so this will just largely be about shellify replacement by shellescape.

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (master@6929634). Click here to learn what that means.
The diff coverage is 23.07%.

@@            Coverage Diff            @@
##             master     #184   +/-   ##
=========================================
  Coverage          ?   20.54%           
=========================================
  Files             ?       27           
  Lines             ?     2565           
  Branches          ?        0           
=========================================
  Hits              ?      527           
  Misses            ?     2038           
  Partials          ?        0

Impacted Files	Coverage Δ
src/funfuzz/js/js_interesting.py	20.34% <100%> (ø)
src/funfuzz/util/subprocesses.py	11.36% <12.5%> (ø)
src/funfuzz/js/compare_jit.py	16.36% <14.28%> (ø)
src/funfuzz/util/lithium_helpers.py	13.72% <20%> (ø)
src/funfuzz/js/compile_shell.py	17.23% <33.33%> (ø)
src/funfuzz/js/inspect_shell.py	22.33% <50%> (ø)

---

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 6929634...b80859f. Read the comment docs.

[nth10sd]

This latest build failure is unrelated. During the process, mozilla-central upgraded its minimium Python requirement to be 3.5+ (see bug 1451065).

All the checks pass at the previous ec0a3a1 changeset here.

[jschwartzentruber]

Don't use str.join([list comprehension]), use str.join(generator expression) instead.
List comprehension builds the whole list in-memory and then calls join() on the result. A generator expression works as though each item is yielded to join() as it is evaluated. The syntax is the same, just drop the square brackets.

This occurs on nearly every line changed, so I haven't added comments on them all.

[jschwartzentruber]

lgtm