please rebase

I don't like the global RwLock, it adds overhead in a hot path and won't work well with multiple electrs instances running in the same process (like we have in the tests). But explicitly forwarding the config option into has_prevout() would be kind of messy too.

I wonder, perhaps it would be cleaner to handle the initial issuance output using a dummy zero-value txo db entry instead?

Also, the CLI parameter is called --initial-issuance-prevout but it actually takes a txid rather than an outpoint. I would rename it to --initial-issuance-txid. (this does follow the INITIAL_ISSUANCE_PREVOUT naming which I originally chose, but we should probably change that too and at least not let it leak into user-visible options.)

Edit: actually --initial-issuance-txid isn't quite right either since this isn't the txid of the issuance tx itself, but rather of the txid spent by its input. Perhaps --initial-issuance-prevtx?