Consider re-enabling RGS

[rolznz]

By using RGS the hubs do not all have to individually request through gossip, this way we do not request so much data from our connected peers.

• network graph need to be reset if RGS is enabled (for hubs without transient network graph) - this is already done by ldk-node
• latest_rgs_snapshot_timestamp need to be reset if RGS is enabled - fix: reset latest_rgs_snapshot_timestamp on startup if using transient network graph #1193

[bumi]

why is requesting data from the peers a problem? rate limits?

[rolznz]

When there is a brief re-connection to a node it seems LDK by default will try to fetch 1 hour of gossip data. With RGS we can fetch snapshots from the RGS server (hopefully small - both because it's a snapshot and RGS data is optimized, and evenly distributed over time to avoid impact)

[rolznz]

I will close this as the required hub changes are complete - we can enable it if/when we are ready