#15061: Expose replicate and 1D shard mappers

[jjiangTT]

Ticket

Link to Github Issue

Problem description

The current pybinds leverage pytorch for sharding, replication, and concatenation rather than doing it natively. This is being fixed slowly with a series of PRs that will deprecate the existing ones in distributed.py with c++ pybinds backed by xtensor.

What's changed

Expose C++ pybinds from existing classes in distribute_tensor.cpp/hpp
Extend distributed/api.cpp's aggregate_tensor to accept borrowedstorage types and aggregate them into a multihoststorage tensor as well
Add temporary forks in ttnn/operations/core.py to redirect cpp classes to the new path as the old one is replaced
Change all existing usages of replicate and shard to

Checklist

• All post commit CI passes
• Blackhole Post commit CI passes (if applicable)
• Model regression CI passes (if applicable)
• Device performance regression CI passes (if applicable)
• (For models and ops writers) Full new models tests CI passes (if applicable)
• New/Existing tests provide coverage for changes: tests/ttnn/distributed/test_distributed_tensor.py