[integer] Implement sqrt() for BigUInt

[forfudan]

This pull request introduces several changes, primarily focusing on adding new functionality, improving debugging, and optimizing arithmetic operations for the BigUInt type. Additionally, it includes updates to documentation and minor refactoring. Below is a summary of the most significant changes:

New Functionality

• Added support for benchmarking the square root operation in benches/biguint/bench.mojo. This includes importing the bench_sqrt function, updating the command options to include sqrt, and adding the corresponding execution logic.
• Introduced the subtract_inplace_by_uint32 function to perform in-place subtraction of a UInt32 value from a BigUInt number. This function includes handling for borrow propagation.

Debugging Enhancements

• Added multiple debug_assert statements across arithmetic operations (add, subtract, multiply, etc.) to check for uninitialized BigUInt values and ensure there are no leading zero words. These assertions improve error detection and debugging during development.

Optimizations and Refactoring

• Replaced slice-based initialization (e.g., BigUInt(words=y.words[bounds_y[0]:bounds_y[1]])) with the more efficient BigUInt.from_slice method in several places, improving clarity and performance.
• Simplified the multiply_slices_karatsuba function by adding a zero-check for input slices, avoiding unnecessary computations when one of the slices is zero.

Documentation Updates

• Updated the docs/todo.md file to reflect completed tasks and clarify the purpose of the to-do list.

Code Cleanup

• Removed unused or redundant functions, such as negative_inplace and absolute_inplace, as they served no meaningful purpose for unsigned integers.