Improve model handling speed

[flferretti]

This pull request includes several code improvements and bug fixes. Here is a summary of the changes:

• Use structural pattern matching (NOTE: This requires Python>=3.10, so I can eventually drop this)
• Use utf-8 encoding when reading from files (Avoids file read error in Windows)
• Cache tree transforms
• Cache gazebo executable path
• General exceptions improvements
• General typing improvements

Benchmarks:

%timeit _ = rod.Sdf.load(sdf=urdf_path, is_urdf=True)

< 683 ms ± 24.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)
> 434 ms ± 11.9 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

%timeit sdf.model.switch_frame_convention(frame_convention=rod.FrameConvention.Urdf)

< 511 ms ± 8.95 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)
> 44 ms ± 178 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

%timeit _ = UrdfExporter.sdf_to_urdf_string(
    sdf=sdf, pretty=True, gazebo_preserve_fixed_joints=True
)

< 549 ms ± 5.71 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)
> 73.1 ms ± 432 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

C.C. @diegoferigo

[diegoferigo]

@flferretti there are many conflicts after #34 and #35. Do you want to port the changes over the future main? Otherwise I can take care of it since I was the one that introduced all the conflicts.

[flferretti]

@flferretti there are many conflicts after #34 and #35. Do you want to port the changes over the future main? Otherwise I can take care of it since I was the one that introduced all the conflicts.

I can handle that, thank you!

[flferretti]

Ready for review @diegoferigo, I also fixed a minor typo here https://github.com/ami-iit/rod/blob/11d276868a6720439d8ce9db9d64f209500a1b79/src/rod/utils/frame_convention.py#L297