Grounding!

No circuit breaker I know of can handle a direct lightning strike. A direct strike can easily produce millions of amperes of current from the millions of volts of potential between cloud and ground. The only thing good about a direct lightning strike is that it is not hard to find what has been damaged. An indirect lightning strike is another thing. Most of the time the induced currents generated by the EMP (Electro Magnetic Pulse) of lightning are small and relatively brief enough that no damage occurs at all. It also helps that the magnitude of the pulse field linearly decreases with the square of the distance from the direct strike. For locations that it is likely for lightning to strike a structure a separate earthing contact from the power distribution is usually required.

The grounding electrode at a generator is needed for the exact same reason as the secondary of the downstream transformer should have a grounded electrode. I circuit breaker attached to a source can only detect a ground fault downstream if the source is grounded. This gets far more complicated with long distance, multiple power source and load conditions of the power grid even though the general principles are the same.

i: The other half of the closed path loop of current path is the much larger path of the distributed capacitor of cloud to ground that the arc is not travelling.

ii: It depends on the wiring convention of your local wiring codes. The NFPA codes for most of the US require power distribution to be earthed and for permanent conductive building structure (metallic plumbing and structural steel) to be separately earthed from power distribution. The NFPA codes recognize that some installations require isolated returns or phases instead of neutral wires. By convention, the definition of a neutral conductor requires a connection to earth. This earth connection wire must only carry fault currents when they happen. A return wire need not be connected to earth.

My point here is that there is no Physics reason for turning return wires into neutral wires by connecting them to earth. However, when electrical wiring fails the installed infrastructure needs a wiring convention followed to attempt the safe mitigation of a fault. These codes came about by methodical studies of disasters where property and lives have been lost. From time to time these codes are reviewed in a tedious process.

Query #1 has been pretty well answered, so I will dwell on #2.

The conductor (call it B) that you refer to as going from the panel (frame of panel I guess you mean) to the transformer would indeed provide protection against live conductor faults to the panel frame or to any other exposed metal that is connected to B, but consider other possibilities where the live conductor comes into contact with structural metal (or any other metal) that is not connected to B. All of that metal is now at live potential (it could be the entire building) and no protective equipment will trip.

If someone concurrently touches that metal and the frame of the switchboard or other B connected devices they will complete the fault circuit and receive a possibly fatal shock. An installed RCD could operate to reduce the shock severity. Connecting that B conductor to common ground by means of a stake will reduce the possibility of unearthed structure and provide a return path for the fault current, and protective devices will operate more reliably and without waiting for someone's body to complete the fault path.

Now consider another possibility:- Resident #1 and #2 live next door to each other or even in the next street or suburb.

Conductor B to resident #1's property is open circuit, he doesn't have an installed RCD, he has a kettle the outer frame of which is live, so, while every item of resident #1's equipment is connected to his internal B conductor, they are not connected to the main B and so their exposed metal frames are effectively at full line potential. His protective devices do not operate to isolate the circuits, and he is unaware of any faults and indeed suffers no ill effects, everything works as it should. The kettle is sitting on the sink which now comes to line potential.

Resident #2 next door (who does have an installed RCD) is about to boil her kettle and as she holds the kettle and grabs the tap handle, the fault current from next door travels through the copper pipes to her tap and through her body to her kettle which IS effectively connected to B. She receives a potentially fatal shock. Her RCD registers no imbalance and does not open to protect her.

The earth stake would have prevented this from occurring by providing an earth reference for resident #1's fault current. Even if his individual earth connection was invalid, the fault current would have been shunted to earth via the copper pipes, operating his protective devices and thus causing no shock to #2

There are numerous other valid reasons for connecting one live conductor to earth but that should be enough to illustrate some safety factors involved.