Only consider tests to be manual tests as a last resort.

[qyearsley-zz]

@gsnedders @jgraham

This is one very simple idea for #196 - would this be OK to do? This would result in many tests being reclassified from "manual" to "testharness" or "reftest".

Context: Previously for Chromium we had some logic that would look at the test file itself and look for "testharness.js" or "" to guess if the test was a testharness or reftest, and not import it if didn't look like either of those, and then run it as a testharness or ref test regardless of what the manifest says. But, it would be better to rely on what the manifest says, so I want to remove that logic.

---

This change is 

[jgraham]

@gsnedders already complained about my much more conservative change in #197 which is believed by inspection not to break anything. So I'm going to say that this change which does (we certainly have manual tests that also use testharness.js) isn't a great idea.

[codecov-io]

Codecov Report

Merging #199 into master will not change coverage.
The diff coverage is 50%.

@@           Coverage Diff           @@
##           master     #199   +/-   ##
=======================================
  Coverage   87.75%   87.75%           
=======================================
  Files          24       24           
  Lines        2418     2418           
  Branches      406      406           
=======================================
  Hits         2122     2122           
  Misses        230      230           
  Partials       66       66

Impacted Files	Coverage Δ
manifest/sourcefile.py	82.54% <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 51ba63c...6ab7cc8. Read the comment docs.

[qyearsley-zz]

Hm, alright :-) thanks for quickly taking a look and making a decision.