test(ci): coverpkg=gno.land/... for txtar tests

[moul]

Note: I'm uncertain about what will happen after the merge.

Fixes #3085
Addresses #3003

[gfanton]

I tried something more generic in 9adfc59 (#3086); unfortunately, it seems to makes the gnovm tests timeout :(

There is a chance that this PR could work since it only applies to gno.land, but having this kind of different rules per module/package will make potential similar issues, like the one you had in #3003, harder to debug in the futur.

---

side note: why is this PR not triggering test jobs?

[moul]

side note: why is this PR not triggering test jobs?

Probably because of 629fef9 :)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 63.74%. Comparing base (d73b6c6) to head (1d282a5).
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3088      +/-   ##
==========================================
+ Coverage   63.33%   63.74%   +0.40%     
==========================================
  Files         548      548              
  Lines       78680    78680              
==========================================
+ Hits        49835    50153     +318     
+ Misses      25481    25143     -338     
- Partials     3364     3384      +20     

Flag	Coverage Δ
contribs/gnodev	60.54% <ø> (-0.63%)	⬇️
contribs/gnofaucet	14.82% <ø> (ø)
gno.land	73.62% <ø> (+6.48%)	⬆️
gnovm	67.90% <ø> (ø)
misc/genstd	79.72% <ø> (ø)
tm2	62.34% <ø> (-0.09%)	⬇️

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.

[moul]

Alternative option: #3090

[thehowl]

Reporting my 2c from signal:

I personally think it's fine if txtar tests aren't counted for coverage. Two reasons:

1. The set up to make coverage work with txtar is ugly (maybe not with gnoland, but IIRC gnokey builds a separate binary and then runs it, so this setup is required there). I've seen what Guillhem had set up and it requires joining coverage files and creating related reports; there's no way to integrate this directly into go test -cover.
2. txtar should be integration tests in addition to other unit tests. They're good for verifying specific conditions and testing out features in "real life" scenarios, but they are slow. Testing for normal features should go in unit tests, so should the verification and coverage of all code paths.

I'd prefer if we kept txtar tests to something similar to #2159, ie. integreation-testing common user flows, and then regression tests for complex bugs and behaviours which are hard to test.

[thehowl]

#3090 (comment)