FIX: CBA_fnc_deleteEntity looping over network and kill dedicated server

[Vdauphin]

When a unit is created inside a group, the group locality isn't immediately the same as the unit. It is 0, which mean the CBA_fnc_deleteEntity is remoteExecCall to every client. So because the unit is only local to server or one client, all others clients will remeteExecCall again the CBA_fnc_deleteEntity to server because groupOwner can only be tested on it. So the remote execution is starting to loop between all client and server, until the groupOwner return a real client (not 0).

Code testing the groupOwner command (executed server side in debug console):

private _group = createGroup west;
_group createUnit ["B_officer_F", [0, 0, 0], [], 0, "CARGO"];
diag_log format ["Locality test without delay: isLocalUnit %1, groupOwner %2", local ((units _group) select 0), groupOwner _group];
//_group call CBA_fnc_deleteEntity;
[{
	params ["_args"];
	diag_log format ["Locality test with delay: isLocalUnit %1, groupOwner %2", local ((units _args) select 0), groupOwner _args];
}, _group, 2] call CBA_fnc_waitAndExecute;

Returned:

12:56:54 "Locality test without delay: isLocalUnit true, groupOwner 0"
12:56:56 "Locality test with delay: isLocalUnit true, groupOwner 2"

This loop can dramatically kill the network (>500kbits) and the dedicated server FPS.

When merged this pull request will:

• To prevent that the CBA_fnc_deleteEntity execution is waiting until the groupOwner is correct (not 0)
• Test if groupOwner isn't 0 before remoteExecCall
• If groupOwner is 0 wait until it isn't. Then _group call CBA_fnc_deleteEntity;
• Bug introduced by this FIX: cba_fnc_deleteEntity where group must be local to the machine executing deleteGroup command #916

[dedmen]

Worst CBA bug this decade ^^

[commy2]

Minor fix. Quit loop should entity become null by other means instead of looping endlessly.