Use routers to fetch routing tables

[madchicken]

This PR aims to use the list of router servers the cluster provides to fetch the routing table for a specific DB.
With these modifications we also fix the fact that when asking for the default DB the driver was not using any connection pool, but it was creating a direct connection to the main cluster URL using the provided configuration (causing the whole TLS handshacking to happen every time).

The main changes introduce the concept of a "router" server pool, update the interfaces and implementations to propagate this information, and add round-robin selection for routers. The changes also update tests and documentation accordingly.

Routing and Connection Management Enhancements:

• Added support for passing and selecting a "router" (ConnectionPool) throughout the routing and connection registry APIs, including in servers, fetch_routing_table, and get_default_db methods. This allows more efficient and flexible routing table fetching and server selection. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12]

• The RoutingTableProvider trait and its ClusterRoutingTableProvider implementation now accept an optional router connection pool, using it to obtain connections for routing table queries. [1] [2] [3] [4]

Load Balancing Improvements:

• Extended the LoadBalancingStrategy trait and RoundRobinStrategy to support router selection, including a new router_index for round-robin router selection. Added the select_router method and logic to filter and sort routers for consistent selection. [1] [2] [3] [4]

Documentation and API Quality:

• Improved documentation for session creation and default database retrieval in the Graph implementation, clarifying caching and session behavior.

Test Updates:

• Updated tests to pass the new router parameter where required and to reflect the updated APIs, ensuring continued coverage and correctness. [1] [2] [3] [4] [5] [6]

[madchicken]

@knutwalker this is now ready to be reviewed, we tested it against a real cluster

[madchicken]

@knutwalker any chance to have this PR merged?

[knutwalker]

@madchicken i'll look at it next week

[knutwalker]

LGTM!
There's a small typo in a doc comment, I'll leave it to you to merge

[knutwalker]

Sorry, just merged #275 and it introduced a conflict with your changes. Once rebased, you should also be getting the now updated CI builds for the status checks