#8364: Disable implicit fallback for ttnn.reshape

[ayerofieiev-tt]

What's happening
This is a part of the #8364 effort

ttnn.rehape now do not support automatic/implicit fallback to torch.
Instead, model developers have to explicitly use torch.
Use ttnn.get_fallback_function(ttnn.reshape) for simplicity.

Update

• Disabled implicit fallback
• Added ttnn.get_fallback_function to make explicit fallback easier
• Updated tests and models to call fallback explicitly where needed

CI
Change checked on CI with:

• Post commit
• Nightly fast dispatch
• T3000 frequent tests
• Model perf regression and output report

There are two remaining failures on CI, which I am ok with:

• performance - need to bump the threshold. Someone might have to investigate perf drop unrelated to this changes (aligned with @tt-rkim)
• Nightly timeouted after 60 minutes - this seem to reproduce in main too