Close incremental decommit spec/impl gaps

[ch1bo]

Why

Incremental decommit specification was updated by #1473 and we have a long-lived feature branch open which implements #1057 according to these spec changes. This item is to reconcile the implementation of the branch with the plan made in the specification, such that we have a working and consistent (maybe imperfect) state to be merged and maybe even released.

What

Clear red areas from #1473 by reviewing their coverage in #1344

• Overall protocol life cycle and off-chain logic changes must be covered by integration and unit tests
• Each change of on-chain code must be covered by a mutation test
• Differences in implementation are to be resolved according to spec (if possible)

[ch1bo]

Most major things I'm addressing of the (many) comments I left on #1344:

• [X] checkClose must not use version from redeemer
• [X] Close/Contest redeemer sum type
• [X] checkLastKnownVersion -> construct and observe correct version
• [X] Observe decrement outputs
• [X] Remove snapshot number check in close
• [x] Client request decommit: Check against local ledger -> inconsistent "user interface" logic
• [x] CloseTx chain layer interface -> remove redundancies / make invalid cases impossible
• [x] Maybe Hash vs. Maybe UTxO -> take a decision

[ch1bo]

With #1513 ready to review, there is now only two things left to do:

• [x] Model versions in TxTrace
• [x] Move redundant / complex e2e tests (corner cases) into lower levels
• [x] Model spec concerns (comments)