Add fileshare process monitoring

[devzbysiu]

Changes:

• JobMonitorFileshareProcess replaced with NetlinkProcessMonitor - based on netlink Process Events Connector API:
  • monitor listens for EXIT and EXEC events from the kernel
  • EXEC event handler verifies that the process created is a fileshare process by getting PID from the event and reading /proc VFS to check the process path
  • if it's confirmed that the created process is a fileshare process, then we save its PID and allow fileshare port
  • on EXIT, we compare the PID and if it does match then we are blocking fileshare because it means that fileshare process exited - no matter how (was killed or stopped)
• the monitor is enabled only when meshnet is on

[mariusSincovici]

What would happen if PIDs are recycled or if the numbering is restarted(when max int was reached) by the kernel ?

[devzbysiu]

Good point. Not sure how likely this it. I'll measure the performance without this if check, because it will read /proc for all new processes then

[devzbysiu]

The Process Event Connector mechanism is not working on docker because of process and user namespaces mismatch and it's not reporting errors to user-space. I'm dropping this approach.