Agent need restart to correctly fill the nodes table

[aletundo]

The package nodestbl.go is used for handling the information about the other nodes in the network, such as the max rate limits and the address and port of the relative HAProxy server instance.

The agent needs to be restarted after the deployment (and I guess also undeployment...) of the functions. Otherwise, the table content is not updated and empty/bad messages are sent to the other peers (since the table entries are invalid or outdated).

N.B.: the nodes table is currently used by our naive load distribution algorithm. In future versions, a new algorithm implementation may not need this package, and this issue can be closed without a fix.