MD improvement on trailing data and long foreign (non-pure latin) data

[Ousret]

Close #121

Address as best as humanly (or bot-wise) possible some remaining edge cases.

Coming from uchardet, those two files acquired my attention.

• https://gitlab.freedesktop.org/uchardet/uchardet/-/blob/master/test/zh/gb18030.txt
• https://gitlab.freedesktop.org/uchardet/uchardet/-/blob/master/test/zh/big5.txt

Both files contain the same content. Repeating in loop "classical chinese" without any punctuation.
They were both miss-detected as cp1251 and cp037.
To be absolutely clear, as they are borderline cases, the goal is not to get it right but be closer.

Now, with this patch, they are both detected as "big5". And they are nothing that we can do without adding considerable effort.
The problematic case is the gb18030 --> big5 that output 潠极笢恅潠极笢恅潠极笢恅潠极笢恅潠极笢恅潠极笢恅潠极笢恅潠极笢恅潠极笢恅潠极笢恅潠极笢恅. And Google Translate seems to think that is translatable to Extremely strong.

Nonetheless, It is a positive PR that increases the detector accuracy.

[codecov-commenter]

Codecov Report

Merging #124 (5772bc3) into master (8b52c35) will increase coverage by 0.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #124      +/-   ##
==========================================
+ Coverage   90.25%   90.27%   +0.01%     
==========================================
  Files          11       11              
  Lines        1160     1162       +2     
==========================================
+ Hits         1047     1049       +2     
  Misses        113      113              

Impacted Files	Coverage Δ
charset_normalizer/md.py	96.65% <100.00%> (+0.02%)	⬆️

---

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 8b52c35...5772bc3. Read the comment docs.