Switch from Coveralls to Codecov

[ryanio]

(Successor to #645)

Due to codecov/codecov-action#29, it does not look like GH Actions can post to CodeCov on PRs from forks due to the inability to share secrets.

CircleCI does not need CODECOV_TOKEN, so this PR tests pushing the coverage reports using CircleCI.

Once codecov/codecov-action#29 is resolved, we can remove the use of CircleCI and re-enable .github/workflows/vm-coverage.yml with the proper fix.

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@e32f732). Click here to learn what that means.
The diff coverage is n/a.

@@           Coverage Diff            @@
##             master    #646   +/-   ##
========================================
  Coverage          ?   91.4%           
========================================
  Files             ?      31           
  Lines             ?    1978           
  Branches          ?     326           
========================================
  Hits              ?    1808           
  Misses            ?      90           
  Partials          ?      80

Flag	Coverage Δ
#vm	91.4% <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 e32f732...e5c9415. Read the comment docs.

[ryanio]

Looks like it's now working well pushing from circleci to codecov without needing a token to be set, so it will work on PRs from forks :) https://codecov.io/gh/ethereumjs/ethereumjs-vm/pulls

@holgerd77 if this looks good to merge, please update the check from ghactions-coverage to circleci-coverage and should be good to go!

[holgerd77]

Just comparing the coverage results from Coveralls (right) and Codecov (left). Coverage is systematically lower on Codecov, I would guess that their metrics work a bit differently and there is a somewhat more strict count in place. Wouldn't make this a blocker, especially in this direction (lower count) this is relatively unproblematic since it doesn't give a false sense of security / good coverage.

[holgerd77]

This is really great, thanks a lot @ryanio! Really looking forward to the new coverage reports, confident that this will have a significant effect on test quality / coverage considerations along with PRs. This went always a bit under radar (unless someone wanted explicitly improve the coverage), since the info was very much inconvenient to consume along the normal work flow. So: cool! 😄

[evertonfraga]

Regarding the lower percentage, although I don’t see it as a problem as well, we can test some other settings to see how it behaves.
https://docs.codecov.io/docs/codecovyml-reference#section-parsers