Add SynthWordGenerator to text reco training scripts

[felixdittrich92]

After some feedback now this PR integrates only the WordGenerator into the training scripts 😄

Any feedback is welcome 😃

[codecov]

Codecov Report

Merging #825 (8403853) into main (41237e9) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main     #825   +/-   ##
=======================================
  Coverage   95.97%   95.97%           
=======================================
  Files         131      131           
  Lines        4988     4988           
=======================================
  Hits         4787     4787           
  Misses        201      201           

Flag	Coverage Δ
unittests	95.97% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
doctr/transforms/modules/base.py	94.59% <0.00%> (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 41237e9...8403853. Read the comment docs.

[fg-mindee]

Thanks a lot Felix!

Would you mind just making this PR about fixing the german vocab please?
And about your suggestion to replace the character by a double S, I think it's a dangerous idea because a vocab element = a theoretical character = a computer character so far. With this, we'd be breaking this rule.

[fg-mindee]

Thanks! For now, it would be good to have a similar transformation set as other recognition tasks (cf. my comments). Let me know what you think 👌

[fg-mindee]

Thanks for the edits! Still a few things to address and we'll be good to go!

[fg-mindee]

Thanks Felix!

Final adjustments to make (my comments also apply to the tensorflow script) :)

[fg-mindee]

Thanks Felix 🙏