[MetaSchedule]Fix the bug when loading database_tuning_record.json if there is pad_einsum primitive

[YXY-0922]

When loading from database_tuning_record.json in Meta Schedule (this line: B_reindex_pad_shared_dyn[v0, v1] = T.if_then_else(v0 < 1, B[v1, v0], T.float16(0))), the parameter dtype of the primitive pad_einsum is read as int64, causing a block iterator v0 that should be int32 to be inferred as int64. This results in an InternalError: Check failed: (ret_ex.dtype() == var.dtype()) is false: substituting v0:int32 -> v0:int64. This commit performs a type conversion within UnpackedApplyToSchedule to fix this bug.

for ax0_ax1_fused in range(2048):
    with T.block("B_reindex_pad_shared.dyn"):
        v0 = T.axis.spatial(16, ax0_ax1_fused // 128)
        v1 = T.axis.spatial(4096, ax2_0_0 * 128 + ax0_ax1_fused % 128)
        T.reads(B[v1, v0])
        T.writes(B_reindex_pad_shared_dyn[v0, v1])
        T.block_attr({"buffer_dim_align": [[0, 0, 32, 8]], "meta_schedule.cooperative_fetch": 1})
        B_reindex_pad_shared_dyn[v0, v1] = T.if_then_else(v0 < 1, B[v1, v0], T.float16(0))

[Hzfengsy]

Sorry for the late response, and thanks for the great catch. One thing is that did you test if the TIR function itself is i64-based? It would be good if you could add test cases