parfor_divide and parfor_exponent does not work on int32 and int64 if device does not support float64

[ZzEeKkAa]

If we try to run parfor_divide or parfor_exponent function on device that does not support f64 with i32 or i64 arguments it will output lowering error.
Occurred on gen12 igpu. Reproducible is inside tests.

================================================================================================================= FAILURES =================================================================================================================
________________________________________________________________________________________ test_built_in_operators1[parfor_exponent-device-int32-100] ________________________________________________________________________________________
self = <numba_dpex.core.parfors.parfor_lowerer.ParforLowerImpl object at 0x7ff400b90280>, lowerer = <numba.parfors.parfor_lowering.ParforLower object at 0x7ff3f00b2a30>
parfor = id=24[LoopNest(index_variable = parfor_index.1004, range = (0, a_size0.997, 1))]{25: <ir.Block at /home/yevhenii/Proje...umba-dpex/numba_dpex/tests/dpjit_tests/parfors/test_builtin_ops.py (35)>}Var(parfor_index.1004, test_builtin_ops.py:35)

    def _lower_parfor_as_kernel(self, lowerer, parfor):
        """Lowers a parfor node created by the dpjit compiler to a
        ``numba_dpex.kernel``.

        The general approach is as follows:

            - The code from the parfor's init block is lowered normally
              in the context of the current function.
            - The body of the parfor is transformed into a kernel function.
            - Dpctl runtime calls to submit the kernel are added.

        """
        # We copy the typemap here because for race condition variable we'll
        # update their type to array so they can be updated by the kernel.
        orig_typemap = lowerer.fndesc.typemap

        # replace original typemap with copy and restore the original at the
        # end.
        lowerer.fndesc.typemap = copy.copy(orig_typemap)

        if config.DEBUG_ARRAY_OPT:
            print("lowerer.fndesc", lowerer.fndesc, type(lowerer.fndesc))

        typemap = lowerer.fndesc.typemap
        varmap = lowerer.varmap

        loc = parfor.init_block.loc
        scope = parfor.init_block.scope

        # Lower the init block of the parfor.
        for instr in parfor.init_block.body:
            lowerer.lower_inst(instr)

        for racevar in parfor.races:
            if racevar not in varmap:
                rvtyp = typemap[racevar]
                rv = ir.Var(scope, racevar, loc)
                lowerer._alloca_var(rv.name, rvtyp)

        alias_map = {}
        arg_aliases = {}

        find_potential_aliases_parfor(
            parfor,
            parfor.params,
            typemap,
            lowerer.func_ir,
            alias_map,
            arg_aliases,
        )

        # run get_parfor_outputs() and get_parfor_reductions() before
        # kernel creation since Jumps are modified so CFG of loop_body
        # dict will become invalid
        if parfor.params is None:
            raise AssertionError

        parfor_output_arrays = get_parfor_outputs(parfor, parfor.params)

        # compile parfor body as a separate dpex kernel function
        flags = copy.copy(parfor.flags)
        flags.error_model = "numpy"

        # Can't get here unless
        # flags.set('auto_parallel', ParallelOptions(True))
        index_var_typ = typemap[parfor.loop_nests[0].index_variable.name]

        # index variables should have the same type, check rest of indices
        for loop_nest in parfor.loop_nests[1:]:
            if typemap[loop_nest.index_variable.name] != index_var_typ:
                raise AssertionError

        loop_ranges = [
            (loop_nest.start, loop_nest.stop, loop_nest.step)
            for loop_nest in parfor.loop_nests
        ]

        parfor_redvars, parfor_reddict = parfor.redvars, parfor.reddict

        nredvars = len(parfor_redvars)
        if nredvars > 0:
            self._reduction_codegen(
                parfor,
                typemap,
                nredvars,
                parfor_redvars,
                parfor_reddict,
                lowerer,
                parfor_output_arrays,
                loop_ranges,
                flags,
                alias_map,
            )
        else:
            try:
>               parfor_kernel = create_kernel_for_parfor(
                    lowerer,
                    parfor,
                    typemap,
                    flags,
                    loop_ranges,
                    bool(alias_map),
                    parfor.races,
                    parfor_output_arrays,
                )

numba_dpex/core/parfors/parfor_lowerer.py:480:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
numba_dpex/core/parfors/kernel_builder.py:440: in create_kernel_for_parfor
    sycl_kernel = _compile_kernel_parfor(
numba_dpex/core/parfors/kernel_builder.py:88: in _compile_kernel_parfor
    kernel_bundle = dpctl_prog.create_program_from_spirv(
dpctl/program/_program.pyx:274: in dpctl.program._program.create_program_from_spirv
    ???
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

>   ???
E   dpctl.program._program.SyclProgramCompilationError

dpctl/program/_program.pyx:311: SyclProgramCompilationError

During handling of the above exception, another exception occurred:

fmt_ = 'lowering "{inst}" at {loc}', args = ()
kwargs = {'inst': id=24[LoopNest(index_variable = parfor_index.1004, range = (0, a_size0.997, 1))]{25: <ir.Block at /home/yevhe...lename=/home/yevhenii/Projects/numba-dpex/numba_dpex/tests/dpjit_tests/parfors/test_builtin_ops.py, line=35, col=None)}
errcls = functools.partial(<class 'numba.core.errors.LoweringError'>, loc=Loc(filename=/home/yevhenii/Projects/numba-dpex/numba_dpex/tests/dpjit_tests/parfors/test_builtin_ops.py, line=35, col=None))
loc = Loc(filename=/home/yevhenii/Projects/numba-dpex/numba_dpex/tests/dpjit_tests/parfors/test_builtin_ops.py, line=35, col=None)
newerr = LoweringError('Failed in dpex_dpjit_nopython mode pipeline (step: lowerer with support for parfor nodes)\n\x1b[1m\x1b[...ps.py:35)" at /home/yevhenii/Projects/numba-dpex/numba_dpex/tests/dpjit_tests/parfors/test_builtin_ops.py (35)\x1b[0m')
tb = None

    @contextlib.contextmanager
    def new_error_context(fmt_, *args, **kwargs):
        """
        A contextmanager that prepend contextual information to any exception
        raised within.  If the exception type is not an instance of NumbaError,
        it will be wrapped into a InternalError.   The exception class can be
        changed by providing a "errcls_" keyword argument with the exception
        constructor.

        The first argument is a message that describes the context.  It can be a
        format string.  If there are additional arguments, it will be used as
        ``fmt_.format(*args, **kwargs)`` to produce the final message string.
        """
        errcls = kwargs.pop('errcls_', InternalError)

        loc = kwargs.get('loc', None)
        if loc is not None and not loc.filename.startswith(_numba_path):
            loc_info.update(kwargs)

        try:
>           yield

../../.miniconda3/envs/rc2025/lib/python3.9/site-packages/numba/core/errors.py:823:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
../../.miniconda3/envs/rc2025/lib/python3.9/site-packages/numba/core/lowering.py:285: in lower_block
    self.lower_inst(inst)
../../.miniconda3/envs/rc2025/lib/python3.9/site-packages/numba/parfors/parfor_lowering.py:51: in lower_inst
    _lower_parfor_parallel(self, inst)
../../.miniconda3/envs/rc2025/lib/python3.9/site-packages/numba/parfors/parfor_lowering.py:60: in _lower_parfor_parallel
    return parfor.lowerer(lowerer, parfor)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Discovered while working on #1143