Filter unused fragments from peak_df

[GeorgWa]

Add remove_unused_peaks method to MSData_Base for efficient peak dataframe cleanup after spectrum filtering. Includes a numba-accelerated mask generator for high performance and supports both in-place and copy operations with full class inheritance preservation.

This PR was 1.87USD

[jalew188]

Any redundencies with alphabase's remove unused fragments?

[github-actions]

The implementation adds functionality to remove unused peaks from MS data, which is useful for memory efficiency. The code is generally well-structured with good documentation. The main issues are:

1. The use of Numba without error handling if Numba is not available (consider adding a fallback implementation)
2. The get_peaks_to_keep_mask function could be more efficient by using vectorized operations instead of loops
3. Consider adding parameter validation to the remove_unused_peaks method to check if the required columns exist in the dataframes
4. The test cases are thorough, but they don't test error cases or edge cases like empty dataframes

[github-actions]

Number of tokens: input_tokens=11268 output_tokens=2153 max_tokens=4096
review_instructions=''
config={}
thinking: ```
[]

[mlorenz49]

Tested this out on test file. Function is working and properly removing peaks from peak_df upon filtering spectrum_df.

[GeorgWa]

Any redundencies with alphabase's remove unused fragments?

Yes, potentially. This is something which could be improved in the future.