fix the CI failure with the RPC HTTP endpoint.

[Ethnical]

The CI is failling multiples times each time because the RPC nodes are rate limited etc and this is adding few errors each time that can be tedious each time and loose interest of the CI.

The idea we had was to use private RPC node (like the superchain registry):

1. take the private RPC node and make a regex that replace the RPC_URL in the .env of the concerned task by the RPC based on the path (sep = sepolia url rpc) and (eth = ethereum url rpc)