Fix issues in Octuple with multiple different-beat time signatures

[ilya16]

Hi! I've been playing around with miditok>3.0.0 and found two bugs in the Octuple encoding:

1. Octuple tokenizer with multiple time signatures with different beat values (e.g., time_signature_range = {2: (1, 2), 4: (1, 4), 8: (1, 8)}) encodes and decodes positions so that they are shorter than in the original MIDI. For example, note positions in */4 bars are decoded twice as short within the bar. The example MIDI in Incorrect Placement of Time Signatures #131 will not be decoded correctly if there is a time signature */8 in the config.
   
   Solution: adapt the position encoding logic from REMI to Octuple.
2. The duration of the first note after the time signature change might be encoded incorrectly.
   Solution: move the processing of time signatures above the note duration.

---

📚 Documentation preview 📚: https://miditok--146.org.readthedocs.build/en/146/

[codecov]

Codecov Report

Attention: 3 lines in your changes are missing coverage. Please review.

Comparison is base (37e28ed) 91.22% compared to head (4d98f7a) 91.01%.
Report is 1 commits behind head on main.

Files	Patch %	Lines
miditok/tokenizations/octuple.py	84.21%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #146      +/-   ##
==========================================
- Coverage   91.22%   91.01%   -0.21%     
==========================================
  Files          33       33              
  Lines        4978     4985       +7     
==========================================
- Hits         4541     4537       -4     
- Misses        437      448      +11     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Natooz]

Nice catch, thank you for the fix(es)!
I'll definitely have to include MIDIs with more complex time configuration in the tests