Handle DRAND API outages

[bajtos]

See e35c089 and https://space-meridian.slack.com/archives/C06RPCL6QGL/p1724249441844879

DRAND provides several API endpoints; see https://drand.love/docs/http-api-reference:

Protocol Labs

• https://api.drand.sh/
• https://api2.drand.sh/
• https://api3.drand.sh/
  Cloudflare
• https://drand.cloudflare.com/
  StorSwift
• https://api.drand.secureweb3.com:6875/

We should implement a fallback to other API endpoints when the primary one is down.

A possibly better & easier alternative is to use FastestNodeClient, see drand-client docs:

// if you're happy to get randomness from many APIs and automatically use the fastest
// you can construct a `FastestNodeClient` with multiple URLs
// note: the randomness beacons are cryptographically verifiable, so as long as you fill
// in the `chainVerificationParams` in the options, you don't need to worry about malicious 
// providers sending you fake randomness!
const urls = [
    'https://api.drand.sh',
    'https://drand.cloudflare.com'
    // ...
]
const fastestNodeClient = new FastestNodeClient(urls, options)
// don't forget to start the client, or it won't periodically optimise for the fastest node!
fastestNodeClient.start()

[juliangruber]

+1 to using FastestNodeClient

[bajtos]

Hmm, after reading the implementation, I don't think FastestNodeClient is what we want. You need to run a benchmark to pick the fastest API provider. Then, the request is sent to this fastest provider only.

I wish drand-client provided something like FallbackProvider in Ethers.js.

[bajtos]

I opened a feature request in drand-client repo: drand/drand-client#84