[docs] Benchmark uses unoptimized EFS mount options

[Atry]

What happened:
here:

juicefs/docs/en/benchmark/mdtest.md

Line 40 in fd191ea

mount -t nfs -o nfsvers=4.1,rsize=1048576,wsize=1048576,hard,timeo=600,retrans=2,noresvport, <EFS-ID>.efs.<REGION>.amazonaws.com:/ /efs

nconnect is not specified, while in Amazon EFS performance tips, nconnect=64 or more is recommended for high IOPS

What you expected to happen:
The mount option should be optimized according to Amazon EFS performance tips

How to reproduce it (as minimally and precisely as possible):
Visit https://juicefs.com/docs/community/mdtest/

Anything else we need to know?
I hope the benchmark could be more convincing

Environment:
N/A

[jiefenghuang]

Thank you for your feedback, we will look into the issue.