Improve multi coordinator support

[elias-garcia]

Improve how we handle POST API requests to multiple coordinators

[elias-garcia]

New edge case detected:

Scenario

We have one coordinator + boot coordinator. The next coordinator to forge is the first one, and the node is down, or even the DNS has not been configured properly and the API is not reachable. The boot coordinator already has the create account authorization.

Bug

When sending the create account authorization to both coordinators, the first one will return a connection refused error and the boot coordinator will return a 409 as it already has the create account authorization. hermezjs will then throw the first error, which is the connection refused one, and the error doesn't even have an status or a response.

Proposal

It would be great to filter this errors from not reachable nodes to avoid throwing in this edge case.