Batch operations

[denesterov]

Add support for batch operations.

For example we need to update 100k records with different keys.
At the moment there is only one option:

• split the data by replicaset and then by bucket_id
• manually call each replicaset, with appropriate data
• on each replicaset update individual buckets, not to forget to call 'vshard.storage.bucket_refrw' / 'bucket_unrefrw'.
• take care of situation when bucket was moved during operation

It would be reasonable to have that functionality in vshard. A call like that:

vshard.batch_write(data, storage_write_func_name)
where 'data' is a map of key -> record and specified storage functions takes single 'data' element as argument and stores it.

and a similar call like that:
vshard.batch_read(data, storage_read_func_name)

[Gerold103]

Fixed in #491.