Too many discards is now also a problem in conformance

[lehins]

Today on CI I ran into this failure:

Failures:

  test/Test/Cardano/Ledger/Conformance/Spec/Conway.hs:27:7: 
  1) Conway.Conformance, Ticks transition graph, RATIFY
       *** Gave up! Passed only 98 tests; 1000 discarded tests:
        0% None
       
       83% CC yes votes ratio  	=0%
        8% CC yes votes ratio  	0%-20%
        1% CC yes votes ratio  	=100%
       
       85% DRep yes votes ratio	=0%
        4% DRep yes votes ratio	=100%
        3% DRep yes votes ratio	0%-20%
       
       53% SPO yes votes ratio 	=0%
       38% SPO yes votes ratio 	0%-20%
        1% SPO yes votes ratio 	40%-60%
       
        3% Accepted at least one NoConfidence

  To rerun use: --match "/Conway/Conformance/Ticks transition graph/RATIFY/" --seed 457525973

This needs fixing otherwise it is gonna nag us forever as the old tracing testing system still does occasionally

[TimSheard]

Does this men we are not generating enough input with enough votes that actually pass?

[lehins]

From the looks of it is that during generation of random data we are doing too many discards and looks like a 1000 discards is a limit.

Maybe @Soupstraw has a better insight into what is actually going on.

[Soupstraw]

I don't know for sure, I haven't changed anything in RATIFY generators. I suspect it might have something to do with the exists constraints that I used to generate more valid votes.

Do we know when these kinds of failures started to appear?