[bug]: minter cannot restart caretaker for batches already broadcast

[jharveyb]

If, during minting, a batch is broadcast but then the caretaker encounters an error, the broadcast batch will exist on disk but the minter will not be aware of TX confirmation.

We have no way to start a new caretaker to manage that batch - we can only start caretakers for the pending batch, but re-adding the same assets would create a different batch (with a different funding input, etc.) Also, if the mint was already broadcast we want to make sure we recognize that minting event. Ideally we could specify a batch on disk, perhaps cancelled or not-cancelled, to finalize instead of just the pending batch.