The idea is to use kernel NIC trigger for LED. The sfpswitch for
TurrisOS has to disable LED functionality.

Support for OpenWRT version of ip utility:
- different path
- different output syntax

Fix typos in seldomly-used method calls.

Add support for user-forced mode in sfpswitch daemon.
Interface has to be restarted ony in case when the interface
is already up and when the mode has really changed.
Rewrite the command execution to subprocess calls.
Rewrite sfpswitch internal logic.
Make sfpswitch read EEPROM and detect particular modules.
Make sfpswitch use locking.

Add support of SGMII SFP modules to the sfpswitch daemon.
The daemon has to read I2C EEPROM in the SFP module to determine module
type. If the module seems to be 1000BASE-T, then set it to SGMII mode.

Refacotr sfpswitch script in order to reuse code.
Add support for better --oneshot (set SFP, do not daemonize)
Add support for --resetled (Reset LED status, do not touch SFP mode,
do not daemonize).

Add LED indication to the sfpswitch daemon for Omnia.
The WAN LED has to be driven by software when SFP is inserted.
LED lights when the LOS and FLT signals are inactive.
The DIS signal has to be set to inactive.

The sfpswitch daemon has to bring the NIC down for sure before
calling network restart.

Add safety timeout to be sure the network down/up happens after
the switch is comitted to the kernel.

Omnia has GPIO expander that generates IRQs when the SFP is inserted or
removed.

mvneta driver in Omnia kernel is hacked to export attribute that sets
the target PHY. Catch is that interface has to be restarted after a switch
of the target PHY has happened.

This daemon connects this together: It waits for IRQ (in fact for GPIO
change) and performs the PHY switch and restart of network.

This is only a temporary solution for the meantime between Omnia release
and phylink driver adoption.