improve baggage performance

[gruebel]

Description

• During my work on stabilizing baggage in the Rust project I checked the implementation in other languages and saw some possible improvements.
• The suggested implementation keeps the original intent by not exposing the mutable baggage dict and creates a copy before modifying it. If you are ok with it, then I can also remove the copy() and directly manipulate the baggage in the context, which would further improve the performance.
• The current approach improves the performance by roughly 10x

you can easily test it

    import timeit
    from types import MappingProxyType

    data = {i: i for i in range(10)}  # if the baggage has more values, then performance gain becomes even bigger

    time_proxy = timeit.timeit(lambda: dict(MappingProxyType(data)), number=100000)
    time_copy = timeit.timeit(lambda: data.copy(), number=100000)

    print(f"proxy: {time_proxy:.6f} seconds")
    print(f"copy: {time_copy:.6f} seconds")
    
    -> proxy: 0.042052 seconds
    -> copy: 0.004569 seconds

Checklist:

• Followed the style guidelines of this project
• Changelogs have been updated
• Unit tests have been added
• Documentation has been updated

[emdneto]

Interesting. I was curious about the results for larger baggage set, and opened #4468 with a benchmark so we can compare the results.

But, in summary the results for baggage size of 10000 were improved by ~8.4x

[xrmx]

@gruebel Please add a changelog entry.

This also shows that we have a very broad typing 😅

[gruebel]

@xrmx done 🙂

[gruebel]

@emdneto @lzchen is anything missing or is it good to merge? 😄

[xrmx]

Wrong release 😢

[gruebel]

Oh, sorry 🙈