Add blank state transitions to state

[dr-orlovsky]

Fixes #80

The bug was caused by the fact that we do state update only on consignment consumation - and not on production of state transitions. This PR fixes that.

[zoedberg]

I've tried this fix on rgb-lib. It solves the blank transition issue, but it breaks some other tests that were previously succeeding. We'll try to reproduce the new bugs with the sandbox so that you can see it yourself

[nicbus]

Since an rgb-lib test fails when the code from this PR is used, using our sandbox I tried to initiate a transfer (up to rgb transfer) but then abort it (don't validate, broadcast the tx nor accept the transfer on either side), then re-start the transfer from scratch (including re-generating the recipient invoice).

Once the first transfer has been initiated, rgb state shows the change allocation from that transfer (instead of the original one), although the related tx was never broadcast.

This situation can be reproduced using this branch of the sandbox.

As an example, given the following initial situation:

• outpoints:

dbfa837f699077039b395a37277b4d160f91c422184714554aeed204f27842b5:0

• allocations:

2000  utxo=dbfa837f699077039b395a37277b4d160f91c422184714554aeed204f27842b5:0

after initiating a transfer (txid 946a3a73598a9ec45033f354596665fd0b3f070ced821a0492a9636c5a15d2b2) this is what I get:

• outpoints:

dbfa837f699077039b395a37277b4d160f91c422184714554aeed204f27842b5:0

• allocations:

amount=1900  utxo=946a3a73598a9ec45033f354596665fd0b3f070ced821a0492a9636c5a15d2b2:0

#82 might also be related.

[dr-orlovsky]

@nicbus I fixed the problem of removing historic state in RGB-WG/rgb-core#166

Please re-test with that fix - I think all tests will be passing now