[FR]: add a new pass MergeModel to use with SplitModel

[xieofxie]

Proposal Summary

For static quantization, some model is too big to run (I will get OOM). So I need to split it first.
But the result models are smaller, so I think we had better have a MergeModel pass to merge them back to one model

What component(s) does this request affect?

• OliveModels
• OliveSystems
• OliveEvaluator
• Metrics
• Engine
• Passes
• Other

[devang-ml]

Are the resulting parts smaller because you optimized/quantized each split before trying to merge them together?

[xieofxie]

Yes, for uint8, usually only 25%