Efficiency improvements in cd/alphabet_languages

[adbar]

• I thought that the accentuation check would be more efficient on a set
• The values for the features target_have_accents & target_pure_latin can be cached so the code is not run all the time

Another idea: if you don't use character position in the most frequent characters per language in assets/ the chars could be a set/frozenset instead of a list, that would make the line [c for c in language_characters if c in characters] faster.

[codecov-commenter]

Codecov Report

Merging #122 (a0ceedd) into master (38cfa45) will increase coverage by 0.04%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #122      +/-   ##
==========================================
+ Coverage   90.30%   90.35%   +0.04%     
==========================================
  Files          11       11              
  Lines        1165     1171       +6     
==========================================
+ Hits         1052     1058       +6     
  Misses        113      113              

Impacted Files	Coverage Δ
charset_normalizer/cd.py	96.49% <100.00%> (+0.08%)	⬆️
charset_normalizer/constant.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 38cfa45...a0ceedd. Read the comment docs.

[Ousret]

I took a quick look over and there are some remarks. Make sure to run the bin/run_autofix.sh to ensure that linters won't fail.
Regarding the "performance" effect/benefit I don't have the time to verify it for now.

[Ousret]

Some final thought on this.

[Ousret]

I brought some minor modifications to your fork.
Well, after several tests. The performance gap between Chardet and Charset-Normalizer is slightly better.
Before x4.861
After x4.916

x0.1 faster.
Repeated the test several times.