No `--wfail` when building archive

Stream: mathlib4

[Daniel Weber]

Is the lack of --wfail when building Archive and Counterexamples intentional? (It's still caught by the "check for noisy stdout lines" CI step)

[Ruben Van de Velde]

Oversight, I think

[Damiano Testa]

Archive and Counterexamples are independent of each other, I think, so building both, even if they fail, can give useful information.

[Michael Rothgang]

#19633 adds --wfail

[Michael Rothgang]

The archive and counterexamples are built as separate steps... ah, but by default the former step failing means the latter will be skipped. Ok, that should be added.

[Damiano Testa]

The end result is not much different, though, right? Both get build and either one failing is flagged.

[Michael Rothgang]

Sure! If adding --wfail makes somebody less puzzled when looking at the file, I consider that an improvement. If adding some if condition to a workflow step is confusing again, perhaps it's less relevant.