tapgarden: minting batch display for pre-0.3.3 batches

[jharveyb]

DB is shared privately.

It seems that in either #827 or #866 , DB query behavior was changed such that listing batches (the ListBatches RPC) now displays assets from a batch in their current state, and not their genesis state from the asset minting.

With the new checks added in #866 where we recompute the anchor output key when fetching managed UTXOs, this breaks batch listing once those assets have moved.

This breaks batch listing once any asset from the batch is spent.

[jharveyb]

New itest to repro the issue:

https://github.com/GeorgeTsagk/taproot-assets/blob/8902b720c5b32fc95537a55de3daadc4dab5bac4/itest/assets_test.go#L532

When fetching a broadcast, finalized, etc. batch, we try to fetch the (sprouted) assets with the DB query FetchAssetsForBatch. This just fetches assets with a genesis ID from the batch. This means that the results include further 'instances' of the asset that result from asset transfers, like asset change. That is incorrect; the expectation for ListBatches is that it would list the assets created in a finalized batch, but not any other asset 'instances'.

To fix this, I think we can update marshalMintingBatch and fetchAssetSprouts s.t. for batches in a state beyond Committed:

• Fetch the seedlings for the batch; these are not updated once the batch is frozen.
• Use the seedlings and batch info to construct fully populated proof Locators for each seedling in the batch.
• Use FetchAssetProofs to fetch the issuance proofs for all assets in the batch.
• Read the assets from the issuance proofs and use those to build a tap commitment, which should match what was created when sprouting the batch.

IIUC we can use proof.SparseDecode to only decode the genesis asset and not the whole proof to avoid some overhead. That decoding could also happen in parallel.