Add data type workaround for binary ops

[mmanzoorTT]

Ticket

Closes #2321

Problem description

tt-metal has introduced type checker for binary ops which aborts program
in case of any mismatch.
tt-metal PR: tenstorrent/tt-metal#17828

What's changed

Add a workaround for binary ops.

Checklist

• New/Existing tests provide coverage for changes

[sdjordjevicTT]

Thanks for the follow up!

[kmabeeTT]

Hey guys @mmanzoorTT / @sdjordjevicTT want some thoughts - tt-metal folks just replied to my question in slack channel #tt-metal-ttnn about whether this assert might be overly tight (given llama was passing for us) and they are planning to remove the assert for now because it regressed passing models. So while we probably don't 100% need this PR, I think it's still worthwhile to keep for now to unblock llama without waiting for newer tt-metal tomorrow, and it can be updated as needed in a while depending on where they land with (potentially updated/less tight) assert, does this sound good?

[mmanzoorTT]

Hey guys @mmanzoorTT / @sdjordjevicTT want some thoughts - tt-metal folks just replied to my question in slack channel #tt-metal-ttnn about whether this assert might be overly tight (given llama was passing for us) and they are planning to remove the assert for now because it regressed passing models. So while we probably don't 100% need this PR, I think it's still worthwhile to keep for now to unblock llama without waiting for newer tt-metal tomorrow, and it can be updated as needed in a while depending on where they land with (potentially updated/less tight) assert, does this sound good?

@kmabeeTT I agree