Version 2.0 - Candidate

[Ousret]

This package is reaching its two years of existence, now is a good time for a nice refresh.

Main improvements:

⚡

• 4x to 5x faster than the v1.4
• 2x faster than chardet v4.0

And up to 10x faster when the "preemptive" detector is active (default active)

• Accent has been made on UTF-8 detection, as it is the most common one nowadays. Should perform rather instant.

🎨

• Code refactoring to ease the readability and maintainability

➖

• Dropping cached_property for Python 3.5

After much consideration, this release won't drop python 3.5 as of yet.
This PR will ship with greater care around the CI/CD. Sorry in advance, but it is going to be strict to ensure backward compatibility.

[Ousret]

Performance tests using GHA runner.
Original claims are now confirmed and solid. On large/medium payload, the performance gap is huge.

On normal-sized files

------------------------------
--> Chardet Conclusions
   --> Avg: 0.08440268115942029s
   --> 99th: 0.61577s
   --> 95th: 0.33059s
   --> 50th: 0.03201s
------------------------------
--> Charset-Normalizer Conclusions
   --> Avg: 0.04982222222222222s
   --> 99th: 0.48197s
   --> 95th: 0.2635s
   --> 50th: 0.01967s

On medium-sized files

Apply x16 on the original file

------------------------------
--> Chardet Conclusions
   --> Avg: 1.1437113768115943s
   --> 99th: 9.66639s
   --> 95th: 4.76477s
   --> 50th: 0.30817s
------------------------------
--> Charset-Normalizer Conclusions
   --> Avg: 0.06866171497584542s
   --> 99th: 0.64688s
   --> 95th: 0.41864s
   --> 50th: 0.02136s

On large-sized files

Apply x32 on the original file

------------------------------
--> Chardet Conclusions
   --> Avg: 2.3235273188405796s
   --> 99th: 20.01236s
   --> 95th: 10.22639s
   --> 50th: 0.62519s
------------------------------
--> Charset-Normalizer Conclusions
   --> Avg: 0.07789222222222222s
   --> 99th: 0.67865s
   --> 95th: 0.45936s
   --> 50th: 0.02218s