Misc qbft test fixes

[Zacholme7]

This pr addresses a few issues that I had while implementing the QBFT spec tests

1. All fulldata was mocked to a [1,2,3,4] byte sequence which we do not allow to be run with qbft. To address this, I moved TestingQBFTFullData to be valid data. Alongside this, the TestingQbftRootData was also fixed. This also applies to the DifferentFullData which I adjusted to be different valid data.

2. Fix the double hash in the create proposal hash for the create message spectests. The root was being passed it which was leading to a double hash so I adjusted it to pass in the fulldata instead

3. Fix ending roots in create message tests. The expected roots were hardcoded in and after I changed the fulldata and hash of fulldata, I also had to adjust the expected ending root for comparison

4. I removed some test

• invalid_val_check.go: this was testing that a decided message should pass validation even if it has invalid data which should not be possible
• invalid_identifier.go & no_instance_running.go: these felt like implementation detail tests. we are initializing new qbft instances for new identifiers that we encounter as long as it passes validation to account for slow BNs

5. adjust round change justifications in create_proposal_previously_not_prepared. the round change justifications were for round 2 but the round is being overridden for 10, so the justifications should be for 10. same applies for create_proposal_previously_prepared

[dknopik]

LGTM