tests involve fragile RNG that fails in some contexts

[ExpandingMan]

I get the following failures on Pkg.test("ReTest")

MiscSeed: Test Failed at /home/expandingman/.julia/dev/ReTest/test/runtests.jl:878
  Expression: MiscSeed.RAND1 === MiscSeed.RAND2 === rands[2]
   Evaluated: 0x3214 === 0x3214 === 0x837e
Stacktrace:
 [1] macro expansion
   @ ~/.julia/dev/ReTest/test/runtests.jl:878 [inlined]
 [2] macro expansion
   @ ~/.julia/dev/ReTest/test/setup.jl:71 [inlined]
 [3] macro expansion
   @ /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/Test/src/Test.jl:1151 [inlined]
 [4] top-level scope
   @ ~/.julia/dev/ReTest/test/setup.jl:71
                  Pass
Main.MiscSeed |      0
MiscSeed: Test Failed at /home/expandingman/.julia/dev/ReTest/test/runtests.jl:880
  Expression: MiscSeed.RAND1 === MiscSeed.RAND2 === rands[2]
   Evaluated: 0x24ae === 0x24ae === 0x837e
Stacktrace:
 [1] macro expansion
   @ ~/.julia/dev/ReTest/test/runtests.jl:880 [inlined]
 [2] macro expansion
   @ ~/.julia/dev/ReTest/test/setup.jl:71 [inlined]
 [3] macro expansion
   @ /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/Test/src/Test.jl:1151 [inlined]
 [4] top-level scope
   @ ~/.julia/dev/ReTest/test/setup.jl:71

However, this doesn't seem to happen in CI/CD, see #41 which has exactly the same tests as I ran to get this but succeeds.

It looks like something needs to be seeded somewhere, but as this seems to be checking something rather tricky I'm not sure where would be appropriate to set the seed.

[rfourquet]

The problem is that we test against different versions of the global RNG, and the hardcoded values correspond to an obsolete version of the RNG on what was nightly, which does not show as failing on CI because of the useless continue-on-error option, which shows all-green although the test actually failed.

[ExpandingMan]

Thanks!