Reftests: factorise some repo hashes

[rjbou]

No description provided.

[kit-ty-kate]

Have you checked that those two tests still have the same behaviours if this patch was applied before their respective fixes?

[rjbou]

conflict-4373 has no core modifications, and list just ec6f65e, and changing repo hash doesn't change the output.

[rjbou]

#5414 used in another test a removed hash in this PR. Need to be reviewed

[kit-ty-kate]

I’m not sure it’s worth potentially breaking carefully crafted tests just to save 3 secondes (i tried) in total test time

[rjbou]

What is broken ?

[kit-ty-kate]

I don’t know, I saying potentially. When writing tests there are many factors that appears when choosing a repository hash, many of which might not appear in the results before or after the commit that added the test and its fix.

To me it’s not worth the hassle.

[rjbou]

In the current state of the PR, I checked that it is the same before and after fix, when there is an after

[kit-ty-kate]

many of which might not appear in the results before or after the commit

as i said, many invariants do not appear in the results before or after

[rjbou]

oups, i misread. Then what are the other things that you check when choosing an hash, except that it permits you to have the "good" test output ?