#12529: Add check for in0_mcast_num_cores=1 for noc_async_write_multicast_loopback_src

[TT-BrianLiu]

Ticket

Link to Github Issue

Problem description

After some recent changes to multicasting in one of the sharded in0 reader kernels, some edge cases (when N=32) are hanging/bad pcc.

What's changed

It seems like noc_async_write_multicast_loopback_src with linking no longer supports the case where we are just doing a local copy from one CB to another. So the fix is to just use noc_async_write for local copying.

• Reverts logic update in 59f969
• Use noc_async_write for local copying between CBs

Checklist

• Post commit CI passes
  • post-commit all: https://github.com/tenstorrent/tt-metal/actions/runs/10924658304
  • post-commit perf: https://github.com/tenstorrent/tt-metal/actions/runs/10924679701
• Blackhole Post commit (if applicable)
• Model regression CI testing passes (if applicable)
• Device performance regression CI testing passes (if applicable): https://github.com/tenstorrent/tt-metal/actions/runs/10924689904
• New/Existing tests provide coverage for changes