fix: mlir error when the split have 1 payload instead of 2

[yadej]

Motivation

A valid schedule crash at the compiler step

import xtc.graphs.xtc.op as O
from xtc.backends.mlir import Backend
from xtc.schedules.descript import descript_scheduler

i, j, k, dtype = 50, 64, 64, "float32"
a = O.tensor((i, k), dtype, name="A")
b = O.tensor((k, j), dtype, name="B")

with O.graph(name="matmul") as gb:
    p = O.matmul(a, b, name="C")
graph = gb.graph

backend1 = Backend(graph)
sch = backend1.get_scheduler()
descript_scheduler(
    scheduler=sch,
    node_name="C",
    abstract_dims=["i", "j", "k"],
    spec={
        "k": {},
        "i": {},
        "i#10":{},
        "i[0:5]" :{
            "i#5": {},
            "j": {},
        },
        "i[5:]" :{
            "i#5": {},
            "j": {},
        }
    }
)
comp = backend1.get_compiler(
    shared_lib=True,
    dump_file="matmul_spec_error",
)
module = comp.compile(sch.schedule())

RuntimeError: MLIR Error: %7 = "transform.structured.split"(%6#0) <{dimension = 0 : i64, static_chunk_sizes = 5 : i64}> : (!transform.any_op) -> !transform.any_op expected to contain 2 payloads but it contains 1 payloads
error: %7 = "transform.structured.split"(%6#0) <{dimension = 0 : i64, static_chunk_sizes = 5 : i64}> : (!transform.any_op) -> !transform.any_op expected to contain 2 payloads but it contains 1 payloads

Solution: adding a fail_on_payload_too_small=False attribute to the split_handle