select count(*) errors with prepared statement and request_plus consistency

https://forums.couchbase.com/t/moved-javasdk-couchbaseserver-n1ql-for-4-1-x-count-problem/8213/12

And my questions are still same:
1) how can i benefit from prepared statement in this case, i.e. how to skip plan preparation phase ?
2) should i use count(a) or count is "smart enough" to pre-optimize query and use only secondary index data to count ? Index = "create index on bucket(a,b) where a is valued".
3) Where can i find docs about returned messages error codes detailed description ?
+ 4) What about this error:

{"msg":"Encoded plan parameter does not match encoded plan of 6d44613366b67887374b910295a09171847a79c1","code":4080}

Repro-sample test:
================
i wrote a simple sync-test (code) below, and multiple runs tells me that there are stable 2 errors for COUNT(1) per each 100 calls for separate run:

0 =>[

{"msg":"Encoded plan parameter does not match encoded plan of 3e5d691a6480801871cae0bffb4814ab34d2befc","code":4080}

]
1 =>[

{"msg":"Encoded plan parameter does not match encoded plan of 3e5d691a6480801871cae0bffb4814ab34d2befc","code":4080}

]
DONE with i = 100
Create 3 indexes like:
"create index i1 on bucket.name(a,b) where a is valued;"
"create index i2 on bucket.name(a,b) where a is valued;"
"create index i3 on bucket.name(a,b) where a is valued;"

You can use even empty buckets. Then run the code :

package n1qlcount;

import com.couchbase.client.java.Bucket;
import com.couchbase.client.java.Cluster;
import com.couchbase.client.java.CouchbaseCluster;
import com.couchbase.client.java.document.json.JsonArray;
import com.couchbase.client.java.env.CouchbaseEnvironment;
import com.couchbase.client.java.env.DefaultCouchbaseEnvironment;
import com.couchbase.client.java.query.N1qlParams;
import com.couchbase.client.java.query.N1qlQuery;
import com.couchbase.client.java.query.N1qlQueryResult;
import com.couchbase.client.java.query.consistency.ScanConsistency;
import java.util.Collections;
import java.util.LinkedList;

public class N1QLCount {

public static void main(String[] args) {
CouchbaseEnvironment ce = DefaultCouchbaseEnvironment.create();
final LinkedList<String> nodes = new LinkedList();
nodes.add("A");
nodes.add("B");
nodes.add("C");
Cluster cluster = CouchbaseCluster.create(ce, nodes);
final String bucketName = "bucket.name";
final String bucketPass = "password";
final String fieldA = "a";
final String fieldB = "b";
Bucket b = cluster.openBucket(bucketName, bucketPass);
final N1qlParams p = N1qlParams
.build()
.adhoc(false)
.consistency(ScanConsistency.REQUEST_PLUS);
final String Q =
"SELECT COUNT(1) from `"
+ bucketName
+ "` WHERE "
+ fieldA
+ "=$1 AND "
+ fieldB
+ "=$2;";
JsonArray ja = null;
N1qlQueryResult r = null;
int count = 100, i = 0;
for(i = 0; i< count; i++) {
ja = JsonArray.empty().add("A" + i).add("B" + i);
r = b.query(N1qlQuery.parameterized(Q, ja , p));
if(r.errors().size() > 0)

{ System.out.println(i + " =>" + r.errors().toString()); }

}
System.out.println("DONE with i = " + i);
b.close();
}
}
So, it's defenetly a bug for 4.1.x. The question is "what we gonna do with this" ?