[FEATURE]: build alias set for repeat blocks and use it to reduce solving time

[YuliangLiu0306]

Describe the feature

The solving time for auto-parallel intra-op solver is unacceptable for LLMs as the number of layers increasing.

We could do following steps to reduce the solving time for LLMs:

• find largest repeated blocks. (already done)

• Use an alias set to force all repeated blocks sharing a same distributed training strategies. When we find a node belong to alias sets and it is not the first element of the set, we do not create a new lpVariable for it, but use the same lpVariable created for the first element of alias set.