count addresses endpionts

[tdroxler]

We can first discussed if we want those endpoints or not. The idea comes from this discussion, we currently could have the number of addresses using the rich-list app, but it might not be the more reliable solution to depend on a third-party app, rather than using our explorer-backend.

We know that counting is slow in PG, but we can use our cache and have some long timeout, like 1 to 5min for such data IMO.

The first endpoint /infos/total-addresses returns every addresses that have at least 1 tx.
The second /infos/total-holders-estimation only return addresses with balance > 0. This is an estimation because if we want the real number, we would need to join the Outputs and Inputs, which is too much expensive. Here we only count the addresses with an empty finalized spent output, but to do the finalization time, we can have a difference.

The current estimation is 92228, so on our website and explorer, we could definitively show: More than 92K holders

there are around 90 addresses more when estimate.

explorer=#       
    SELECT count(DISTINCT outputs.address)
      FROM outputs
        LEFT JOIN inputs
        ON outputs.key = inputs.output_ref_key
        AND inputs.main_chain = true
      WHERE outputs.spent_finalized IS NULL
        AND outputs.main_chain = true
        AND inputs.block_hash IS NULL;
 count 
-------
 92133
(1 row)

explorer=#       
     SELECT count(DISTINCT outputs.address)
      FROM outputs
      WHERE outputs.spent_finalized IS NULL
        AND outputs.main_chain = true
;
 count 
-------
 92228
(1 row)