Summary

This PR overhauls the index manipulation API in src/tensors/indexmanipulations.jl to match TensorOperations dispatch conventions, reduces code duplication in the implementation, and adds a dedicated documentation page.
The goal was a bunch of code simplification, (overall number of lines reduced, even though I added some docs 🎉 )

API changes

• Unified in-place interface: permute!, braid!, transpose!, and repartition! now directly accept α, β, backend, and allocator as optional arguments (with defaults One(), Zero(), DefaultBackend(), DefaultAllocator()), following the TensorOperations dispatch pattern. The old add_permute!, add_braid!, and add_transpose! are deprecated and forward to the new functions.
• allocator support: previously, the index manipulation functions did not support a custom allocator at all. It is now a positional argument in both the public and internal interfaces, consistent with TensorOperations convention.
• Out-of-place functions (permute, braid, transpose, repartition) gain backend as a new keyword argument alongside the now-supported allocator keyword.

Implementation changes

• All permutation operations now route through braid!, eliminating duplicate codepaths.
• Dedicated braid! method added for AdjointTensorMap.
• add_transform! kernels for TensorMap refactored to operate on the raw data vector rather than the full TensorMap. Because the data vector has no symmetry type, this avoids recompilation for every TensorMap type combination, improving compilation time.
• Various minor simplifications throughout.