Introduce new special tokens to seq2seq format

[Taka008]

No description provided.

[codecov]

Codecov Report

Merging #188 (34ddade) into dev (2695580) will increase coverage by 0.13%.
The diff coverage is 96.80%.

@@            Coverage Diff             @@
##              dev     #188      +/-   ##
==========================================
+ Coverage   90.48%   90.61%   +0.13%     
==========================================
  Files          62       62              
  Lines        4897     4957      +60     
==========================================
+ Hits         4431     4492      +61     
+ Misses        466      465       -1     

Files Changed	Coverage Δ
src/kwja/modules/seq2seq.py	97.10% <85.71%> (-1.37%)	⬇️
src/kwja/utils/seq2seq_format.py	94.31% <94.25%> (-5.69%)	⬇️
src/kwja/modules/components/logits_processor.py	99.22% <98.97%> (-0.78%)	⬇️
src/kwja/callbacks/seq2seq_module_writer.py	91.66% <100.00%> (-0.93%)	⬇️
src/kwja/datamodule/datasets/seq2seq.py	88.88% <100.00%> (+0.75%)	⬆️
src/kwja/datamodule/datasets/seq2seq_inference.py	100.00% <100.00%> (ø)
src/kwja/datamodule/examples/seq2seq.py	100.00% <100.00%> (ø)
src/kwja/utils/constants.py	97.43% <100.00%> (+0.29%)	⬆️

... and 3 files with indirect coverage changes