Internal services should not spam the same errors in the logs on every iteration

[Serpentian]

A lot of internal vshard services make network requests. If the remote node is down or disabled, then the number of logs starts to grow very fast and the storage for that logs is not infinite. If the error code is the same, as on prev iteration, then the error message should not be printed:

{"code":77,"base_type":"ClientError","type":"ClientError","message":"connect, called on fd 31, aka <addr>: Connection refused","trace":[{"file":".\/tarantool\/src\/box\/lua\/net_box.c","line":2276}]}
Aug 12 13:51:09 test-fix.i[8556]: main/145/vshard.discovery._static_router/bin.test-fix.i.vshard.replicaset replicaset.lua:689 E> Exception during calling 'vshard.storage.buckets_discovery' on 'uri:3045(admin@addr)': connect, called on fd 19, aka <addr>: Connection refused

Aug 12 13:51:19 test-fix.i[8556]: main/148/shared_storage:sandbox_update_worker/bin.test-fix.i.vshard.replicaset replicaset.lua:689 E> Exception during calling 'vshard.storage.bucket_stat' on 'uri:3043(admin@<addr>)': {"reason":"storage is disabled explicitly","code":33,"type":"ShardingError","message":"Storage is disabled: storage is disabled explicitly","name":"STORAGE_IS_DISABLED"}

Router services: discovery, master_search, failover
Storage services: recover, rebalancer, GC, master_search