pallet-balances: Clean up empty reserves & do not send the same event twice

[bkchr]

When reserves are cleared, the corresponding storage should be cleaned up as well. So, we do not run into MaxReserves with only empty reserves. This pull request also fixes the double sending of the same event when slashing a reserved amount.

[bkchr]

/cmd prdoc --audience runtime_dev --bump patch

[dhirajs0]

Small asymmetry: the beneficiary try_insert isn't guarded, so when repatriate_reserved moves nothing (actual == 0 under Polite, e.g. reserved funds frozen) we insert a zero-amount entry on the receiver — the same leak this PR fixes on the slashed side, just mirrored. One-line guard (if !actual.is_zero()) keeps it symmetric. Low priority since the only in-tree caller uses Status::Free. Intentional?

Failing repro (red on this branch, green with the guard)

#[test]
fn repatriate_reserved_named_reserved_status_does_not_leak_empty_beneficiary_entry() {
	ExtBuilder::default().build_and_execute_with(|| {
		let _ = Balances::deposit_creating(&1, 111);
		let _ = Balances::deposit_creating(&2, 100);

		// Account 1 holds a named reserve `Foo = 10` (free 101, reserved 10).
		assert_ok!(Balances::reserve_named(&TestId::Foo, &1, 10));
		assert_eq!(Balances::reserved_balance_named(&TestId::Foo, &1), 10);

		// Freeze the whole balance so the reserved funds are unmovable 
		Balances::set_lock(ID_1, &1, 111, WithdrawReasons::all());

		// Beneficiary 2 has no entry under `Foo`, so the insert path is taken.
		assert!(Balances::reserves(&2).is_empty());

		// Nothing can move, so `actual == 0` and the call returns the full value.
		assert_eq!(
			Balances::repatriate_reserved_named(&TestId::Foo, &1, &2, 10, Reserved).unwrap(),
			10,
		);
		assert_eq!(Balances::reserved_balance_named(&TestId::Foo, &2), 0);

		// Since nothing moved, the beneficiary must NOT gain a named-reserve entry.
		assert!(
			Balances::reserves(&2).is_empty(),
			"no balance moved, so the beneficiary must not gain a named-reserve entry",
		);

		// No `MaxReserves` (= 2) slot was leaked, so two real reserves still fit.
		assert_ok!(Balances::reserve_named(&TestId::Bar, &2, 5));
		assert_ok!(Balances::reserve_named(&TestId::Baz, &2, 5));
	});
}

I have pushed this as a branch (repatriate-reserved-named-empty-entry-repro) if you'd rather run it directly.

[ggwpez]

Could be another let else since the Err(_) match is trivial.

[ggwpez]

nit: could be a match to hard-error if a variant is ever added to Status.