Use fastest compression level

[dolfinus]

Changes

Fixes #1028

According to this microbenchmark, compression level=1 is the fastest one, then higher values gives max 1-3% better compression levels but much slower

Inspired by:
python/cpython#91349 (comment)

microbenchmark.py

import gzip
import io
import pathlib
import time

import cramjam
from typing_extensions import Buffer


def gzip_compress(data: Buffer, level: int):
    buf = io.BytesIO()
    with gzip.GzipFile(fileobj=buf, mode="w", compresslevel=level) as gzipper:
        gzipper.write(data)
    return buf.getvalue()

def lz4_compress(data, level: int):
    compressor = cramjam.lz4.Compressor(
        level=level, content_checksum=False, block_linked=False
    )
    compressor.compress(data)
    return bytes(compressor.finish())


def zstd_compress(data, level: int):
    return bytes(cramjam.zstd.compress(data, level=level))

data = pathlib.Path('numpy.json').read_bytes()  # https://pypi.org/pypi/numpy/json
size = len(data)

compressor_and_level = []
compressor_and_level.extend([(gzip_compress, level) for level in range(0, 10)])
compressor_and_level.extend([(lz4_compress, level) for level in range(0, 10)])
compressor_and_level.extend([(zstd_compress, level) for level in range(0, 22)])

for compressor, level in compressor_and_level:
    start = time.perf_counter()
    compressed = compressor(data, level)
    end = time.perf_counter()
    elapsed = end - start
    print(f"impl={compressor.__name__} level={level} {len(data)//1024}kB->{len(compressed)//1024}kB in {elapsed:.3f} seconds {len(data)//1048576/elapsed:.2f} MB/s {len(compressed)/len(data)*100.0:.2f}% size")

Tested on CPython 3.13.0

impl=gzip_compress level=0 2476kB->2477kB in 0.007 seconds 291.86 MB/s 100.01% size
impl=gzip_compress level=1 2476kB->556kB in 0.027 seconds 73.88 MB/s 22.47% size
impl=gzip_compress level=2 2476kB->545kB in 0.037 seconds 54.18 MB/s 22.00% size
impl=gzip_compress level=3 2476kB->539kB in 0.064 seconds 31.37 MB/s 21.80% size  # 0.7% better than level=1, 40% slower
impl=gzip_compress level=4 2476kB->507kB in 0.058 seconds 34.40 MB/s 20.48% size
impl=gzip_compress level=5 2476kB->497kB in 0.085 seconds 23.44 MB/s 20.09% size
impl=gzip_compress level=6 2476kB->488kB in 0.065 seconds 30.74 MB/s 19.71% size  # 2.7% better than level=1, 2.4 times slower
impl=gzip_compress level=7 2476kB->485kB in 0.078 seconds 25.55 MB/s 19.60% size
impl=gzip_compress level=8 2476kB->481kB in 0.084 seconds 23.73 MB/s 19.43% size
impl=gzip_compress level=9 2476kB->481kB in 0.067 seconds 29.65 MB/s 19.43% size  # 3% better than level=1, 2.5 times slower

impl=lz4_compress level=0 2476kB->949kB in 0.003 seconds 632.23 MB/s 38.32% size
impl=lz4_compress level=1 2476kB->949kB in 0.003 seconds 634.48 MB/s 38.32% size
impl=lz4_compress level=2 2476kB->931kB in 0.009 seconds 226.64 MB/s 37.63% size  # 0.7% better than level=1, 2.8 times slower 
impl=lz4_compress level=3 2476kB->908kB in 0.014 seconds 146.69 MB/s 36.69% size  # 1.6% better than level=1, 4.3 times slower
impl=lz4_compress level=4 2476kB->907kB in 0.015 seconds 136.24 MB/s 36.66% size
impl=lz4_compress level=5 2476kB->907kB in 0.015 seconds 131.52 MB/s 36.64% size
impl=lz4_compress level=6 2476kB->907kB in 0.015 seconds 130.69 MB/s 36.64% size
impl=lz4_compress level=7 2476kB->907kB in 0.015 seconds 132.31 MB/s 36.64% size
impl=lz4_compress level=8 2476kB->907kB in 0.015 seconds 130.46 MB/s 36.64% size
impl=lz4_compress level=9 2476kB->907kB in 0.016 seconds 124.11 MB/s 36.64% size  # 1.7% better than level=1, 5.1 times slower 

impl=zstd_compress level=0 2476kB->466kB in 0.015 seconds 130.90 MB/s 18.83% size
impl=zstd_compress level=1 2476kB->426kB in 0.008 seconds 266.39 MB/s 17.22% size
impl=zstd_compress level=2 2476kB->436kB in 0.009 seconds 215.26 MB/s 17.61% size, # 0.4% worse than level=1, 20% slower 
impl=zstd_compress level=3 2476kB->466kB in 0.015 seconds 134.14 MB/s 18.83% size  # 1.6% worse than level=1, 2 times slower
impl=zstd_compress level=4 2476kB->477kB in 0.021 seconds 96.24 MB/s 19.30% size
impl=zstd_compress level=5 2476kB->468kB in 0.034 seconds 58.54 MB/s 18.91% size
impl=zstd_compress level=6 2476kB->456kB in 0.041 seconds 48.96 MB/s 18.45% size
impl=zstd_compress level=7 2476kB->456kB in 0.063 seconds 31.66 MB/s 18.45% size
impl=zstd_compress level=8 2476kB->450kB in 0.055 seconds 36.20 MB/s 18.20% size
impl=zstd_compress level=9 2476kB->451kB in 0.059 seconds 33.99 MB/s 18.22% size
impl=zstd_compress level=10 2476kB->450kB in 0.062 seconds 32.49 MB/s 18.19% size
impl=zstd_compress level=11 2476kB->450kB in 0.063 seconds 31.56 MB/s 18.19% size
impl=zstd_compress level=12 2476kB->450kB in 0.083 seconds 24.17 MB/s 18.19% size
impl=zstd_compress level=13 2476kB->450kB in 0.172 seconds 11.61 MB/s 18.19% size
impl=zstd_compress level=14 2476kB->450kB in 0.185 seconds 10.79 MB/s 18.17% size
impl=zstd_compress level=15 2476kB->449kB in 0.207 seconds 9.64 MB/s 18.16% size
impl=zstd_compress level=16 2476kB->407kB in 0.355 seconds 5.64 MB/s 16.45% size
impl=zstd_compress level=17 2476kB->401kB in 0.596 seconds 3.36 MB/s 16.21% size
impl=zstd_compress level=18 2476kB->408kB in 0.776 seconds 2.58 MB/s 16.48% size
impl=zstd_compress level=19 2476kB->396kB in 1.004 seconds 1.99 MB/s 15.99% size
impl=zstd_compress level=20 2476kB->396kB in 1.397 seconds 1.43 MB/s 15.99% size
impl=zstd_compress level=21 2476kB->396kB in 1.547 seconds 1.29 MB/s 15.99% size,  # 1.2%% better than level=1, 206 times slower

Checklist

• I think the code is well written
• Unit tests for the changes exist
• Documentation reflects the changes
• Add a new news fragment into the CHANGES folder
  • name it <issue_id>.<type> (e.g. 588.bugfix)
  • if you don't have an issue_id change it to the pr id after creating the PR
  • ensure type is one of the following:
    • .feature: Signifying a new feature.
    • .bugfix: Signifying a bug fix.
    • .doc: Signifying a documentation improvement.
    • .removal: Signifying a deprecation or removal of public API.
    • .misc: A ticket has been closed, but it is not of interest to users.
  • Make sure to use full sentences with correct case and punctuation, for example: Fix issue with non-ascii contents in doctest text files.

[mikmatko]

@ods Could this be merged? Thanks!