[Feature] Support Python native `and` and `or` expressions in PTODSL

[liggest]

Summary

PTODSL RuntimeValue 目前不能直接用于 Python 原生 and、or 布尔表达式
编译时会尝试把左侧条件使用 bool(...) 转换，引发报错

希望能够支持在 PTODSL 中正常使用 and、or 布尔表达式，并保持短路行为：

• x and y => x 为假时不对 y 求值
• x or y => x 为真时不对 y 求值

and 简单样例

from ptodsl import pto, scalar


@pto.jit(name="and_kernel", kernel_kind="vector", target="a5")
def kernel(value: pto.i32, divisor: pto.i32, out: pto.ptr(pto.i8, "gm")):
    pred = (divisor != 0) and ((value // divisor) > 0)
    scalar.store(pred, out, 0)


kernel.compile()

or 简单样例

from ptodsl import pto, scalar


@pto.jit(name="or_kernel", kernel_kind="vector", target="a5")
def kernel(value: pto.i32, divisor: pto.i32, out: pto.ptr(pto.i8, "gm")):
    pred = (divisor == 0) or ((value // divisor) > 0)
    scalar.store(pred, out, 0)


kernel.compile()

两种情况都会在编译阶段报错：

ptodsl._diagnostics.PTODSLTracingMisuseError: native Python if/while condition cannot consume a PTODSL runtime value during tracing. This value is a device-side SSA/runtime-metadata value, not a Python bool/int. Use pto.if_(...) or pto.for_(...) for device-side control flow, or keep the bound/condition in pto.const_expr.

Motivation / use case

支持 Python 原生 and、or 布尔表达式，能帮助用户更方便地在 PTODSL 中表达复杂一些的条件语句，这在实际算子逻辑中时不时会出现

同时，布尔表达式的短路行为能在不满足左侧（x）条件时减少对右侧（y）无意义的求值，在右侧计算开销较大，或右侧只在左侧满足时才有效的场景中有助于提升计算效率

Proposed API / behavior

No response

Alternatives considered

No response

Additional context

No response