Use Istanbul instrumentation for unittestcli code coverage

[calixteman]

No description provided.

[calixteman]

It's slightly faster with Istanbul and it aligns with unittest.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 56.18%. Comparing base (08eca52) to head (47f0bdc).

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #21189      +/-   ##
==========================================
- Coverage   63.94%   56.18%   -7.76%     
==========================================
  Files         213      213              
  Lines      134597    67223   -67374     
  Branches        0    14519   +14519     
==========================================
- Hits        86064    37771   -48293     
+ Misses      48533    29452   -19081     

Flag	Coverage Δ
fonttest	8.66% <ø> (?)
unittest	56.29% <ø> (-0.02%)	⬇️
unittestcli	54.89% <ø> (-8.44%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[timvandermeij]

Looks good! It's more consistent between test targets, slightly faster and now results in similar numbers between unittestcli and unittest. The numbers also seem more realistic because looking at e.g. https://app.codecov.io/gh/mozilla/pdf.js/pull/21189/blob/src/core/brotli_stream.js?dropdown=coverage it's now correctly reported as 0% covered because we don't have any unit tests for this, while previously it'd be almost 40% (see https://app.codecov.io/gh/mozilla/pdf.js/blob/master/src%2Fcore%2Fbrotli_stream.js) purely because we imported the file somewhere and "accidentally" hit comments/imports and the like. I thus think the new state represents the real functional coverage better too.

Feel free to merge this if @Snuffleupagus agrees too.

[Snuffleupagus]

Maybe we should just skip running coverage in Windows/Chrome, since that combination appears to be slower and thus it's failing intermittently.

r=me, thanks.