proxy: a configurable rate-limiter per endpoint

[hieblmi]

This PR introduces supports per-endpoint rate limiting using a token bucket based on golang.org/x/time/rate.
Each endpoint configuration applies to each L402 individually.

I tested locally with different loop clients, as well as added unit and integration tests.

Example configuration for one endpoint:

    ratelimits:
      - pathregex: '^/looprpc.StaticAddressServer/ServerWithdrawDeposits.*$'
        requests: 2
        per: 1m
        burst: 5

• With 2 requests per minute, a new token arrives every 30 seconds. Burst 5 means the bucket can hold up to 5 tokens and starts full after being idle.
• If you make 2 requests within 10 seconds, you’ve only used 2 of the 5 tokens, leaving 3 tokens. So your next request (and up to 3 more) can go through immediately.
• Once you’ve consumed all 5 tokens, you must wait for refills: one token every 30 seconds (so the next allowed request would be 30s after the last one that emptied the bucket, then another 30s for the next, etc.).
• If you stop sending requests, the bucket refills at the same rate up to the burst capacity (max 5 tokens).

[lightninglabs-deploy]

@starius: review reminder
@Roasbeef: review reminder
@bhandras: review reminder