:egg: Do host to device copy when 'cuda' flag is enabled
Ensure fair comparison between nvTIFF which reads to CUDA memory (even though it's the fastest already), and the GDAL LiberTIFF and image-tiff backend readers.
:triangular_flag_on_post: Gate python bindings behind pyo3 feature flag
More clearly separating pure Rust parts from the pyo3/python bindings to workaround some pyo3 compiliation issues on rustc 1.92.0-beta.1 (3b4dd9bf1 2025-10-28), plus it speeds up compilation when just working on the Rust side.
Re-enabled nightly Rust tests disabled in 939b7b0c8f3e465b45a71cf5c0ff460667bc6f4e/#57.
:beers: Symlink nvtiff.h to targets/x86_64-linux/lib and set RUSTFLAGS
Brute force symlinking to get nvtiff-sys to compile with conda-forge's libnvtiff that is under $CONDA_PREFIX/include/ instead of $CONDA_PREFIX/targets/x86_64-linux/include/ where most other header files are. One key part is to use RUSTFLAGS instead of LD_LIBRARY_PATH to actually get rustc to search the correct lib/ folder for the .so files.