Add prometheus metrics for DHT internals

[shyba]

The SDK already supports exporting metrics over Prometheus if the user wishes to do so. This PR adds the following metrics under the dht_node namespace:

data store

• stored_blobs_x_bytes_colliding: Number of blobs with at least X bytes colliding with this node id prefix
  • This is used to measure the ratio of content on network vs on blockchain. Check Measure local ratio of content on blockchain vs announced on DHT #3521 and the dashboard repo, but it also needs Add sd_hash param to claim_search #3511 merged.
• stored_blobs: Number of blobs announced by other peers
• storing_peers: Number of peers storing blobs announced to this node

protocol

• peer_manager_keys: Number of keys tracked by PeerManager dicts (sum)
  • If that grows too large, we can get memory leaks
• request_sent: Number of requests send from DHT RPC protocol
• request_success: Number of successful requests
• request_error: Number of errors returned from request to other peers
• response_time: Response times of DHT RPC requests
• received_request: Number of received DHT RPC requests

routing table

• peers_in_routing_table: Number of peers on routing table
• peer_x_bit_colliding: Number of peers with at least X bits colliding with this node id
  • This is used for estimating network size. Details on Understand and/or design system on how to measure disk space available #3463, but the raw calculations will live on the dashboard repo. This also tells how close we are from the peers we added on the routing table, which can help dissecating a live node health into how well we are able to find and and add neighbors.
• buckets_in_routing_table: Number of buckets on routing table
  • Debugging purposes.

New endpoints:

When starting the seed node script with --metrics_port=<some_port>, a web server will expose the following endpoints:

• /metrics: Prometheus endpoint with the above metrics.

If desired, routing table peers and announced blobs can be exported in CSV format using --enable_csv_export. However, you should use a firewall and avoid exposing that publicly as the announced blobs list can reach multiple thousands items, easily bringing your server down by malicious users.

• /peers.csv: All peers from the routing table in CSV format (ip, port, node_id)
• /blobs.csv: All blobs announced to the node in CSV format (hex encoded blob hash)

[coveralls]

Coverage remained the same at 69.714% when pulling f689a97 on fix_script into e6efc1a on master.

[lyoshenka]

• drop *_flight counter
• rebase
• ready to merge