Move elements count to its own procedure (blocks, txs and blobs)

[PabloCastellano]

Counting elements in Postgres is costly, and we are currently counting the elements in /blocks, /blobs and /txs, which causes the table data to take some seconds to load.

The first step to tackle this unresponsiveness is to split the current api calls so that the count query is handled by a different API endpoint. This will allow the table to load faster and display the total count when the data is ready.