EVM/Evaluation: rehabilitate stack size

Context

Part. of #7752 (closed).

While generating the filler files for Dencun, we were able to get more precise informations on the expected results. Unfortunately this showed discrepancies for approximately 100 tests that are making use of the stack depth beyong 256 (which is our hack-ish custom limit). I removed the limit on the evaluator size.

Another approach would be to disable all those tests but in my opinion we should'nt as they are testing stuff that we support they just make use of the full stack size which creates slight discrepancies only on the balances that we previously didn't catch as the fillers weren't as complete.

Testing

CI