Fix advance indexing bug and zeros_like sbp bug

[wyg1997]

修复在跑 swin-transformer eager consistent 2d sbp 时遇到的两个问题：

1. AdvanceIndexing 在多维 placement 时，sbp 设置错误；
2. zeros_like op（ones_like 也一样） 结果的 sbp 和输入的 sbp 不一致（这里是由于它走了 sbp 推导的流程，而 B -> ? 的代价都是 0，所以可能会拿到错误的 sbp），重写了这个 UserOp 的 NdSbpInferFn 就可以了。

[github-actions]

CI failed when running job: cuda-module. PR label automerge has been removed

[github-actions]

CI failed when running job: cuda-speed-test. PR label automerge has been removed

[github-actions]

CI failed when running job: cuda-module. PR label automerge has been removed

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

OneFlow resnet50 time: 136.6ms (= 13661.7ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 139.8ms (= 13977.5ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.02 (= 139.8ms / 136.6ms)

OneFlow resnet50 time: 78.5ms (= 7846.8ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 85.5ms (= 8547.7ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.09 (= 85.5ms / 78.5ms)

OneFlow resnet50 time: 52.2ms (= 10445.4ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 57.6ms (= 11515.0ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.10 (= 57.6ms / 52.2ms)

OneFlow resnet50 time: 39.7ms (= 7945.6ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 48.1ms (= 9626.6ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.21 (= 48.1ms / 39.7ms)

OneFlow resnet50 time: 42.2ms (= 8446.8ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 39.4ms (= 7880.6ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 0.93 (= 39.4ms / 42.2ms)

OneFlow resnet50 time: 149.8ms (= 14977.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 160.2ms (= 16020.3ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.07 (= 160.2ms / 149.8ms)

OneFlow resnet50 time: 89.7ms (= 8965.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 101.7ms (= 10165.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 101.7ms / 89.7ms)

OneFlow resnet50 time: 64.7ms (= 12931.4ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 71.6ms (= 14327.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.11 (= 71.6ms / 64.7ms)

OneFlow resnet50 time: 57.1ms (= 11412.5ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 61.6ms (= 12325.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.08 (= 61.6ms / 57.1ms)

OneFlow resnet50 time: 53.9ms (= 10770.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 58.1ms (= 11620.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.08 (= 58.1ms / 53.9ms)