bug: snowflake algorithm of the request-id plugin will generate massive logs if etcd down

[lingsamuel]

Current Behavior

The error check doesn't cover the etcd down case. If etcd is unavailable, every request will retry thousands of times.
Also, in the request-id use case, the snowflake algorithm introduces an unnecessary (and probably shouldn't) dependency on etcd. `uuid' is good enough for this case. We are planning to remove it.

Steps to Reproduce

1. run APISIX with request-id snowflake enabled
2. create route with request-id plugin and snowflake as algorithm
3. disable etcd
4. request to the route

Environment

• APISIX version (run apisix version): 3.3.0