tt-wavelet

One-level FP32 lifting wavelet transforms (LWT) on Wormhole and Blackhole, shipped as a TTNN-linked op library plus standalone `lwt`, `ilwt`, `lwt_2d`, and `ilwt_2d` binaries and a benchmark harness. Builds the whole local stack — TT-Metal, the TTNN Python bindings, and ttnn-wavelet — against the TT-Metal revision pinned in its submodule.