(a) M1, the meter in the loop, is the ammeter; it measures the current flowing through the coil. M2, the meter across the coil, is the voltmeter; it measures the potential difference across the coil.
The rule follows from what each quantity means. Current is what passes through a component, so the ammeter has to be placed in the path and let that same current pass through it. Potential difference is measured between two points, so the voltmeter has to be connected to both of them, across the component.
(b) Both rules exist for the same reason: a meter must not change the thing it is measuring.
The ammeter is in the main circuit, so its own resistance adds to the total resistance of that circuit. Any resistance it adds would reduce the current — the very quantity it was put there to read. Making its resistance nearly zero means the current it reports is, to a good approximation, the current that was flowing before it was connected. An ideal ammeter has zero resistance.
The voltmeter is connected across the component, so it offers the current an alternative path. If its resistance were low it would draw a substantial current through itself, which would reduce the current through the coil and change the potential difference it was trying to measure. Making its resistance very high means almost no current is diverted. An ideal voltmeter has infinite resistance.
(c) Connecting the ammeter across the coil places a near-zero resistance directly in parallel with it — which short-circuits the coil.
The current now has a choice between the coil's resistance and the ammeter's almost nothing, and it takes the easy path. From
I=RVwith R≈0
a very large current flows through the ammeter, limited only by the battery and the rest of the circuit.
Why this destroys the meter: the ammeter's coil is wound from fine wire chosen to carry only a small current. The surge overheats it by I2Rt and can burn it out, bend the needle hard against its stop, or damage the moving mechanism. The battery may also be damaged by the heavy current drawn from it. The reading obtained is meaningless in any case.
The voltmeter connected wrongly is the harmless mistake. Put a voltmeter in series by accident and its huge resistance simply chokes the circuit off: almost no current flows, the coil does nothing, and the readings are useless — but nothing is damaged. An ammeter in parallel is the one that destroys equipment, which is why it is worth checking twice before switching on.