FUSE client resilliency

[onlyjob]

With replication 001 and two volume servers I'm rsync'ing data to FUSE mount.
Then I'm adding a third volume server and after some time stopping second volume server which knocks off rsync with the following error:

rsync: close failed on "{{FILENAME}}": Input/output error (5)
rsync error: error in file IO (code 11) at receiver.c(853) [receiver=3.1.3]

weed mount logs the following:

I0408 19:11:20 73058 upload_content.go:196] failing to upload to http://192.168.0.2:9334/1,3ba72cb0d0cf Post "http://192.168.0.2:9334/1,3ba72cb0d0cf": dial tcp 192.168.0.2:9334: connect: connection refused
I0408 19:11:20 73058 dirty_page.go:179] upload data {{FILENAME}} to http://192.168.0.2:9334/1,3ba72cb0d0cf: Post "http://192.168.0.2:9334/1,3ba72cb0d0cf": dial tcp 192.168.0.2:9334: connect: connection refused
I0408 19:11:20 73058 dirty_page.go:133] {{FILENAME}} saveToStorage [0,25719): upload data: Post "http://192.168.0.2:9334/1,3ba72cb0d0cf": dial tcp 192.168.0.2:9334: connect: connection refused
E0408 19:11:20 73058 filehandle.go:169] flush {{FILENAME}}: upload data: Post "http://192.168.0.2:9334/1,3ba72cb0d0cf": dial tcp 192.168.0.2:9334: connect: connection refused

FUSE client should have gracefully used available volume servers. There were enough volume servers to continue. Client-side failure was unexpected.

[chrislusf]

Added a fix bb78ab9 to retry in case master assigned the request to the wrong volume servers.