sys/ppc64: Spec does uses `VADL::s/umod` without zero check

[Jozott00]

The PPC64 specification uses the VADL::s/umod built-ins without a zero check for the divisor:

openvadl/sys/ppc64/ppc64.vadl

Lines 965 to 970 in d054cd1

	model XFormModInstr (x : AsmOpExt, type : Id) : IsaDefs = {
	instruction AsId($x.asm) : XFormA =
	X(rs) := VADL::$x.op(XW(ra), XW(rb)) as $type
	encoding AsId($x.asm) = {opcd = 0b011111, extopcd = $x.ext, rc = 0b0}
	assembly AsId($x.asm) = ($x.asm, " ", decimal(rs), ",", decimal(ra), ",", decimal(rb))
	}

Because the VADL::s/umod operations are undefined for 0, this causes a crash on the x86 runners, as on x86 a division by zero causes an exception.
https://github.com/OpenVADL/openvadl/actions/runs/22714792101#summary-65861829474

The solution is to check the divisor against 0 as it is done for the division instructions

openvadl/sys/ppc64/ppc64.vadl

Lines 1240 to 1252 in d054cd1

	model XOFormDivInstrExt (r : XOFormDivInstrRec, oe : Bool, rc : Bool) : IsaDefs = {
	instruction AsId($r.x.asm, $orId($oe; $rc)) : XOForm =
	if XW(rb) != 0
	then let res, flags = VADL::$r.x.op($r.a.arg1, $r.a.arg2) in
	let udf = $r.udf in
	let result = if udf then $r.ures else res as Word as $r.type in {
	X(rs) := result
	$SetXERAndCRWithOV3264($oe; false; $rc; mov; udf | flags.overflow)
	}
	else let result = $r.zres as UDWord in {
	X(rs) := result
	$SetXERAndCRWithOV3264($oe; false; $rc; mov; true)
	}